Loading data

# Loading files -----------------------------------------------------------
#loading tidy phyloseq object
phyloseq_unfiltered <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/Phyloseq/PHY_20230521_MGK_host_tidy.rds")

v_phyloseq <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/Phyloseq/PHY_20240212_MGK_host_marker_magu_Jessica.rds")

sample_data <- sample_data(phyloseq_unfiltered$phyloseq_count)

#Metagenote description was made as below
sample_data %>% 
        data.frame(check.names = F) %>% 
        dplyr::select(c("baylor_id", "baylor_other_id", "sample_type", "treated")) %>%
        mutate(nephele_description =
          case_when(
                  sample_type == "Mock" &
                    treated == 0 ~ "Control (untreated) positive mock community",
                  sample_type == "Mock" &
                    treated == 1 ~ "host DNA depleted positive mock community",
                  sample_type == "Neg." &
                    treated == 0 ~ "Control (untreated) negative reagent only controls",
                  sample_type == "Neg." &
                    treated == 1 ~ "host DNA depleted negative reagent only controls",
                  sample_type == "BAL" &
                    treated == 0 ~ "Control (untreated) BAL",
                  sample_type == "BAL" &
                    treated == 1 ~ "host DNA depleted BAL",
                  sample_type == "Nasal" &
                    treated == 0 ~ "Control (untreated) nasal swabs",
                  sample_type == "Nasal" &
                    treated == 1 ~ "host DNA depleted nasal swabs",
                  sample_type == "Sputum" &
                    treated == 0 ~ "Control (untreated) sputum",
                  sample_type == "Sputum" &
                    treated == 1 ~ "host DNA depleted sputum"
                  )
          ) %>%
        group_by(nephele_description) %>%
        mutate(rownum = row_number(nephele_description)) %>%
        mutate(nephele_description_numbered = 
                        paste(nephele_description,
                              rownum,
                              sep = "_")
                ) %>%
        write.csv("Project_SICAS2_microbiome/5_Scripts/MGK/Host_depletion_git/data/LOG_20230913_MGK_HOST_nephele_samplegroup.csv")

Tinyvamp results

# Loading files -----------------------------------------------------------
#loading tidy phyloseq object


tinyvamp_untreated <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/HOST_tinyvamp_decontaminated_Amy_Willis/20230807_amy/untreated_p_hats_all_v3.RDS") %>%
        t()

tinyvamp_lypma <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/HOST_tinyvamp_decontaminated_Amy_Willis/20230807_amy/lypma_p_hats_all_v3.RDS") %>%
        t()

tinyvamp_benzonase <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/HOST_tinyvamp_decontaminated_Amy_Willis/20230807_amy/benzonase_p_hats_all_v3.RDS") %>%
        t()

tinyvamp_host_zero <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/HOST_tinyvamp_decontaminated_Amy_Willis/20230807_amy/hostzero_p_hats_all_v3.RDS") %>%
        t()

tinyvamp_molysis <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/HOST_tinyvamp_decontaminated_Amy_Willis/20230807_amy/molysis_p_hats_all_v3.RDS") %>% 
        t()

tinyvamp_qiaamp <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/HOST_tinyvamp_decontaminated_Amy_Willis/20230807_amy/qiaamp_p_hats_all_v3.RDS") %>% 
        t()

        
#tinyvamp_benzonase <- read_rds("Project_SICAS2_microbiome/4_Data/2_Tidy/HOST_tinyvamp_decontaminated_Amy_Willis/benzonase_p_hats.RDS")

sample_data_tv <- phyloseq_unfiltered$phyloseq_rel %>%
        subset_samples(sample_type %in% c("BAL", "Nasal", "Sputum")) %>%
        sample_data %>% data.frame() %>%
        mutate(treatment = case_when(treatment == "MolYsis" ~ "Molysis",
                                     treatment == "HostZERO" ~ "Host zero",
                                     .default = treatment)) %>%
        mutate("names"= paste(.$original_sample, .$treatment, sep = "_")) %>%
        remove_rownames() %>% column_to_rownames("names") %>% sample_data()

phyloseq_tv <- merge_phyloseq(
        merge_phyloseq(sample_data_tv,
                       otu_table(tinyvamp_untreated,
                                 taxa_are_rows = T)),
        merge_phyloseq(sample_data_tv,
                       otu_table(tinyvamp_lypma,
                                 taxa_are_rows = T)),
        merge_phyloseq(sample_data_tv,
                       otu_table(tinyvamp_benzonase,
                                 taxa_are_rows = T)),
        merge_phyloseq(sample_data_tv,
                       otu_table(tinyvamp_host_zero,
                                 taxa_are_rows = T)),
        merge_phyloseq(sample_data_tv,
                       otu_table(tinyvamp_molysis,
                                 taxa_are_rows = T)),
        merge_phyloseq(sample_data_tv,
                       otu_table(tinyvamp_qiaamp,
                                 taxa_are_rows = T)),
        tax_table(phyloseq_unfiltered$phyloseq_rel)
) 

sample_data(phyloseq_tv)$treatment <-
        phyloseq_tv %>%
        sample_data() %>%
        data.frame(check.names = F) %>% 
        mutate(treatment =
                       case_when(
                               treatment == 
                                       "Molysis"~ 
                                       "MolYsis",
                               treatment == 
                                       "Host zero" ~
                                       "HostZERO", 
                               .default = treatment)) %>%
        .$treatment %>%
        factor(., levels = c("Untreated", 
                             "lyPMA",
                             "Benzonase",
                             "HostZERO",
                             "MolYsis",
                             "QIAamp"))

Alpha diversity indices

alpha_diversity <- function(data) {
        otu_table <- otu_table(data) #%>% .[, colSums(.) !=0]
        S.obs <- rowSums(t(otu_table) != 0)
        sample_data <- sample_data(data)
        data_evenness <- vegan::diversity(t(otu_table)) / log(vegan::specnumber(t(otu_table))) # calculate evenness index using vegan package
        data_shannon <- vegan::diversity(t(otu_table), index = "shannon") # calculate Shannon index using vegan package
        data_hill <- exp(data_shannon)                           # calculate Hills index
        data_dominance <- microbiome::dominance(otu_table, index = "all", rank = 1, aggregate = TRUE) # dominance (Berger-Parker index), etc.
        data_invsimpson <- vegan::diversity(t(otu_table), index = "invsimpson")                          # calculate Shannon index using vegan package
        alpha_diversity <- cbind(S.obs, data_shannon, data_hill, data_invsimpson, data_evenness,data_dominance) # combine all indices in one data table
        sample_data <- merge(data.frame(sample_data), alpha_diversity, by = 0, all = T) %>% column_to_rownames(var = "Row.names")
}
sample_data(v_phyloseq$viral_rel) <- sample_data(alpha_diversity(v_phyloseq$viral_rel))

sample_data(phyloseq_unfiltered$phyloseq_count)$V.obs <- 
        sample_data(v_phyloseq$viral_rel) %>% 
        data.frame %>% 
        dplyr::select("S.obs") %>% 
#        mutate(S.obs = case_when(is.na(S.obs) ~ 0,
#                                 .default = S.obs)) %>%
        .[sample_names(phyloseq_unfiltered$phyloseq_count),]

sample_data(phyloseq_unfiltered$phyloseq_rel)$V.obs <- 
        sample_data(v_phyloseq$viral_rel) %>% 
        data.frame %>% 
        dplyr::select("S.obs") %>% 
#        mutate(S.obs = case_when(is.na(S.obs) ~ 0,
#                                 .default = S.obs)) %>%
        .[sample_names(phyloseq_unfiltered$phyloseq_count),]

sample_data(phyloseq_unfiltered$phyloseq_path_rpk)$V.obs <- 
        sample_data(v_phyloseq$viral_rel) %>% 
        data.frame %>% 
        dplyr::select("S.obs") %>% 
#        mutate(S.obs = case_when(is.na(S.obs) ~ 0,
#                                 .default = S.obs)) %>%
        .[sample_names(phyloseq_unfiltered$phyloseq_count),]

phyloseq <- phyloseq_unfiltered



alpha_diversity <- function(data) {
        otu_table <- otu_table(data) # %>% .[, colSums(.) !=0]
        S.obs <- rowSums(t(otu_table) != 0)
        sample_data <- sample_data(data)
        data_evenness <- vegan::diversity(t(otu_table)) / log(vegan::specnumber(t(otu_table))) # calculate evenness index using vegan package
        data_shannon <- vegan::diversity(t(otu_table), index = "shannon") # calculate Shannon index using vegan package
        data_hill <- exp(data_shannon)                           # calculate Hills index
        data_dominance <- microbiome::dominance(otu_table, index = "all", rank = 1, aggregate = TRUE) # dominance (Berger-Parker index), etc.
        data_invsimpson <- vegan::diversity(t(otu_table), index = "invsimpson")                          # calculate Shannon index using vegan package
        alpha_diversity <- cbind(S.obs, data_shannon, data_hill, data_invsimpson, data_evenness,data_dominance) # combine all indices in one data table
        sample_data <- merge(data.frame(sample_data), alpha_diversity, by = 0, all = T) %>% column_to_rownames(var = "Row.names")
}



sample_data(phyloseq_unfiltered$phyloseq_rel) <- sample_data(alpha_diversity(phyloseq_unfiltered$phyloseq_rel))
sample_data(phyloseq_unfiltered$phyloseq_count) <- sample_data(alpha_diversity(phyloseq_unfiltered$phyloseq_count)) 
sample_data(phyloseq_unfiltered$phyloseq_path_rpk) <- sample_data(alpha_diversity(phyloseq_unfiltered$phyloseq_path_rpk))  

Fig. S1. qPCR figure

Figure S1. Host depletion effects measured by qPCR. (A) total DNA (16S bacterial DNA + human), (B) host DNA, (C) bacterial DNA, and (D) proportion of host DNA.

#1A: Change in total DNA (qPCR)

fS1a <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = log10(DNA_host_nondil + DNA_bac_nondil))) +
        geom_jitter(aes(color = treatment, x = treatment), lwd = 0.2,
                    alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        facet_wrap(~sample_type, scale = "free_x") +
        ylab("log<sub>10</sub>(qPCR total DNA)<br>(ng/μL)") +
        labs(tag = "A") +
        guides(fill = guide_legend(nrow = 1))



#1B: Change in human DNA (qPCR)
fS1b <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = log10(DNA_host_nondil))) +
        geom_jitter(aes(col = treatment, x = treatment), 
                    lwd = 0.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        facet_wrap(~sample_type, scale = "free_x") +
        ylab("log<sub>10</sub>(qPCR host DNA)<br>(ng/μL)") +
        labs(tag = "B")

#1C: Change in 16S DNA (qPCR)
fS1c <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = log10(DNA_bac_nondil))) +
        geom_jitter(aes(col = treatment, x = treatment),
                    lwd = 0.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        facet_wrap(~sample_type, scale = "free_x") +
        ylab("log<sub>10</sub>(qPCR bacterial DNA)<br>(ng/μL)") +
        labs(tag = "C")
        
#1D. Change in % host (qPCR)
fS1d <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = host_proportion)) +
        geom_jitter(aes(col = treatment, x = treatment),
                    lwd = 0.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        facet_wrap(~sample_type, scale = "free_x") +
        ylab("Host DNA ratio") +
        labs(tag = "D")


#output for markdown
figureS1 <- ggarrange(fS1a, fS1b, fS1c, fS1d, common.legend = T , align = "hv")

figureS1 

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS1.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 170, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

figureS1
# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Fig. S2. numbers

Peggy’s comment on 20240912:

I think you need to distinguish between the host depletion negative (stratified by each method ie lyPMA, benzonase, etc) and sequencing negative controls.

revise Table C12 in your R2R so that besides the row for negative control, also stratify by the type of negative control

create a stacked barplot where you facet by type of negative control

Table updated- negative controls stratified by sample type

table1_mock_and_neg <- sample_data(phyloseq$phyloseq_count) %>% data.frame() %>% 
        #dplyr::filter(sample_type %in% c("Sputum", "Nasal", "BAL")) %>% 
        # mutate(S.obs = case_when(S.obs==0 ~ NA,
        #                          .default = S.obs),
        #        F.obs = case_when(F.obs==0 ~ NA,
        #                          .default = F.obs),
        #        V.obs = case_when(V.obs==0 ~ NA,
        #                          .default = V.obs)) %>% 
        mutate(sample_type = case_when(sample_type == "Neg." & treatment == "Untreated" ~ "Neg. (extraction)",
                                       sample_type == "Neg." & treatment == "lyPMA" ~ "Neg. (lyPMA)",
                                       sample_type == "Neg." & treatment == "Benzonase" ~ "Neg. (Benzonase)",
                                       sample_type == "Neg." & treatment == "HostZERO" ~ "Neg. (HostZERO)",
                                       sample_type == "Neg." & treatment == "MolYsis" ~ "Neg. (MolYsis)",
                                       sample_type == "Neg." & treatment == "QIAamp" ~ "Neg. (QIAamp)",
                                       .default = sample_type),
               sample_type = factor(sample_type, 
                                    levels = c("Neg. (extraction)",
                                               "Neg. (lyPMA)",
                                               "Neg. (Benzonase)",
                                               "Neg. (HostZERO)",
                                               "Neg. (MolYsis)",
                                               "Neg. (QIAamp)",
                                               "Mock",
                                               "BAL",
                                               "Nasal",
                                               "Sputum"))) %>%
        group_by (sample_type) %>%
        summarise(`N` = n(),
            #      `Total DNA <br>ng/µL` = paste(format(round(median(picogreen_ng_ul),2), nsmall = 2, big.mark = ","), "<br>(", format(round(quantile(picogreen_ng_ul, 0.25),2), nsmall = 2, big.mark = ","), ", ", format(round(quantile(picogreen_ng_ul, 0.75),2), nsmall = 2, big.mark = ","), ")", sep = ""),
               `Human DNA <br>pg/µL` = paste(format(round(median(DNA_host_ng_uL*1000),1), nsmall = 1, big.mark = ","), " (", format(round(quantile(DNA_host_ng_uL*1000, 0.25),1), nsmall = 1, big.mark = ","), ", ", format(round(quantile(DNA_host_ng_uL*1000, 0.75),1), nsmall = 1, big.mark = ","), ")", sep = ""),
               `Bacterial DNA <br>pg/µL` = paste(format(round(median(DNA_bac_ng_uL*1000),1), nsmall = 1, big.mark = ","), " (", format(round(quantile(DNA_bac_ng_uL*1000, 0.25),1), nsmall = 1, big.mark = ","), ", ", format(round(quantile(DNA_bac_ng_uL*1000, 0.75),1), nsmall = 1, big.mark = ","), ")", sep = ""),
              `QC'd reads<br>reads x 10<sup>6</sup>` = paste(format(round(median(Reads_after_trim/1000000),1), nsmall = 1, big.mark = ","), " (", format(round(quantile(Reads_after_trim/1000000, 0.25),1), nsmall = 1, big.mark = ","), ", ", format(round(quantile(Reads_after_trim/1000000, 0.75),1), nsmall = 1, big.mark = ","), ")", sep = ""),
              `Host reads<br>%` = paste(format(round(median(sequencing_host_prop*100),1),
                                                nsmall = 1, big.mark = ","),
                                         " (",
                                         format(round(quantile(sequencing_host_prop * 100,
                                                               0.25),
                                                      1),
                                                nsmall = 1,
                                                big.mark = ","), 
                                         ", ", 
                                         format(round(quantile(sequencing_host_prop * 100, 0.75),1), 
                                                nsmall = 1, 
                                                big.mark = ","), 
                                         ")", 
                                         sep = ""),
             `Final reads<br>reads x 10<sup>6</sup>` = paste(format(round(median(Final_reads/1000000,
                                                                                 na.rm = T),
                                                                          1),
                                                                    nsmall = 1, big.mark = ","),
                                                             " (", 
                                                             format(round(quantile(Final_reads/1000000,
                                                                                            0.25,
                                                                                            na.rm = T),
                                                                                   1),
                                                                             nsmall = 1,
                                                                             big.mark = ","),
                                                             ", ",
                                                             format(round(quantile(Final_reads/1000000, 
                                                                                   0.75, 
                                                                                   na.rm = T),
                                                                          1), 
                                                                    nsmall = 1, 
                                                                    big.mark = ","), 
                                                             ")",
                                                             sep = "")
              
            ) %>% 
        data.frame(check.names = F) %>% 
        arrange(sample_type) %>%
        rename(`Sample` = sample_type) %>%
        mutate_all(linebreak) %>% kbl(format = "html", escape = F) %>% kable_styling(full_width = 0, html_font = "sans")

table1_mock_and_neg

save_kable(table1_mock_and_neg, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/table1_mock_neg.html", self_contained = T)

Fig. S2. sequencing results

Fig. S2. (A) final reads after removing low quality reads and host-mapped reads and (B) sum of MetaPhlAn mapped reads by sample type.

sample_data %>% data.frame %>% subset(., .$sample_type == "Neg.")

#how were the samples failed in library prep?
sample_data %>%
        data.frame() %>%
        mutate(sample_type = factor(.$sample_type, 
                                    labels = c("Negative control", "Mock", "BAL", "Nasal", "Sputum")))

figS2_raw_reads <- sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>% otu_table %>% colSums()) %>%
        mutate(sample_type = factor(.$sample_type, 
                                    labels = c("Negative control", "Mock", "BAL", "Nasal", "Sputum"))) %>%
        ggplot(aes(x = reorder(baylor_other_id, -Raw_reads),
                               y = Raw_reads/1000000,
                               col = sample_type)) +
                geom_point() +
                theme_classic(base_family = "sans") +
                theme(axis.title.y = element_markdown(), axis.text.x = element_blank()) +
                ylab("Raw reads X 10<sup>6</sup>") +
                xlab("Samples") +
        guides(col=guide_legend(title="Sample type")) +
        scale_color_brewer(type = "qual", palette = 6) +
        labs(tag = "A") #+
        #ylim(c(0, 8.5))

figS2_raw_reads_log <- sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>% otu_table %>% colSums()) %>%
        mutate(sample_type = factor(.$sample_type, 
                                    labels = c("Negative control", "Mock", "BAL", "Nasal", "Sputum"))) %>%
        ggplot(aes(x = reorder(baylor_other_id, -Raw_reads),
                               y = log10(Raw_reads+1),
                               col = sample_type)) +
                geom_point() +
                theme_classic(base_family = "sans") +
                theme(axis.title.y = element_markdown(), axis.text.x = element_blank()) +
                ylab("log<sub>10</sub>(raw reads)") +
                xlab("Samples") +
        guides(col=guide_legend(title="Sample type")) +
        scale_color_brewer(type = "qual", palette = 6) +
        labs(tag = "A") #+
        #ylim(c(0, 8.5))

figS2_final_reads <- sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>% otu_table %>% colSums()) %>%
        mutate(sample_type = factor(.$sample_type, 
                                    labels = c("Negative control", "Mock", "BAL", "Nasal", "Sputum"))) %>%
        ggplot(aes(x = reorder(baylor_other_id, -Final_reads),
                               y = Final_reads/1000000,
                               col = sample_type)) +
                geom_point() +
                theme_classic(base_family = "sans") +
                theme(axis.title.y = element_markdown(), axis.text.x = element_blank()) +
                ylab("Final reads X 10<sup>6</sup>") +
                xlab("Samples") +
        guides(col=guide_legend(title="Sample type")) +
        scale_color_brewer(type = "qual", palette = 6) +
        labs(tag = "B") #+
        #ylim(c(0, 8.5))


figS2_final_reads_log <- sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>% otu_table %>% colSums()) %>%
        mutate(sample_type = factor(.$sample_type, 
                                    labels = c("Negative control", "Mock", "BAL", "Nasal", "Sputum"))) %>%
        ggplot(aes(x = reorder(baylor_other_id, -Final_reads),
                               y = log10(Final_reads + 1),
                               col = sample_type)) +
                geom_point() +
                theme_classic(base_family = "sans") +
                theme(axis.title.y = element_markdown(), axis.text.x = element_blank()) +
                ylab("log<sub>10</sub>(final reads)") +
                xlab("Samples") +
        guides(col=guide_legend(title="Sample type")) +
        scale_color_brewer(type = "qual", palette = 6) +
        labs(tag = "B")


figS2_total_reads <- sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>% otu_table %>% colSums()) %>%
        mutate(sample_type = factor(.$sample_type, 
                                    labels = c("Negative control", "Mock", "BAL", "Nasal", "Sputum"))) %>%
        ggplot(aes(x = reorder(baylor_other_id, -total_read),
                               y = total_read/1000000,
                               col = sample_type)) +
                geom_point() +
                theme_classic(base_family = "sans") +
                theme(axis.title.y = element_markdown(), axis.text.x = element_blank()) +
                ylab("Microbial reads X 10<sup>6</sup>") +
                xlab("Samples") +
        guides(col=guide_legend(title="Sample type")) +
        scale_color_brewer(type = "qual", palette = 6) +
        labs(tag = "C")


figS2_total_reads_log <- sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>% otu_table %>% colSums()) %>%
        mutate(sample_type = factor(.$sample_type, 
                                    labels = c("Negative control", "Mock", "BAL", "Nasal", "Sputum"))) %>%
        ggplot(aes(x = reorder(baylor_other_id, -total_read),
                               y = log10(total_read + 1),
                               col = sample_type)) +
                geom_point() +
                theme_classic(base_family = "sans") +
                theme(axis.title.y = element_markdown(), axis.text.x = element_blank()) +
                ylab("log<sub>10</sub>(Microbial reads)") +
                xlab("Samples") +
        guides(col=guide_legend(title="Sample type")) +
        scale_color_brewer(type = "qual", palette = 6) +
        labs(tag = "C") 


figS2 <- ggarrange(figS2_raw_reads, figS2_final_reads, figS2_total_reads, ncol = 1, nrow = 3, common.legend = T)


figS2_log <- ggarrange(figS2_raw_reads_log, figS2_final_reads_log, figS2_total_reads_log, ncol = 1, nrow = 3, common.legend = T)

figS2

figS2_log

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS2_revised.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 210, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

figS2
# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS2_revised_log.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 210, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

figS2_log
# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Failure crosstab

sample_data <- sample_data(phyloseq_unfiltered$phyloseq_count) %>% 
          data.frame() %>%
          mutate(
              sample_sums = sample_sums(phyloseq_unfiltered$phyloseq_count),
              lib_failed = case_match(lib_failed, TRUE ~ 1, FALSE ~0),
              seq_failed = ifelse(sample_sums == 0, 1, 0),
              lib_seq_failed = ifelse(lib_failed ==1 | seq_failed == 1, 1, 0),
              prop_host = round((Host_mapped/Reads_after_trim), 1)
              )
        xtabs(lib_failed ~ sample_type + treatment, data = sample_data)

##            treatment
## sample_type Untreated lyPMA Benzonase HostZERO MolYsis QIAamp
##      Neg.           0     0         0        0       0      0
##      Mock           0     0         0        0       0      0
##      BAL            0     1         0        1       1      0
##      Nasal          0     4         0        2       4      0
##      Sputum         0     0         0        0       0      0

        xtabs(seq_failed ~ sample_type + treatment, data = sample_data)

##            treatment
## sample_type Untreated lyPMA Benzonase HostZERO MolYsis QIAamp
##      Neg.           0     0         0        0       0      0
##      Mock           0     0         0        0       0      0
##      BAL            1     1         0        0       0      0
##      Nasal          0     0         0        0       0      0
##      Sputum         0     0         0        0       0      0

        xtabs(lib_seq_failed ~ sample_type + treatment, data = sample_data)

##            treatment
## sample_type Untreated lyPMA Benzonase HostZERO MolYsis QIAamp
##      Neg.           0     0         0        0       0      0
##      Mock           0     0         0        0       0      0
##      BAL            1     2         0        1       1      0
##      Nasal          0     4         0        2       4      0
##      Sputum         0     0         0        0       0      0

Library failure - OR (all samples)

Stratified OR cannot be calculated, as some samples showed 0 library failure.

Effect size, standard error (SE) and t-value at a statistical test on library prep failure rate using generalized linear mixed effect model. glm ( sequencing fail ~ sample_type + treatment + sample_type * treatment + (1|subject_id) )

gm1 <- glmer(lib_failed ~ sample_type + treatment + sample_type * treatment + (1|subject_id),
      data = sample_data %>% 
              subset(sample_type %in% c("BAL", "Nasal", "Sputum")) %>% 
              data.frame(check.names = F),
      family = binomial)

gm1 %>% summary %>% .$coefficients

##                                         Estimate  Std. Error       z value
## (Intercept)                          -21.7185928    20795.51 -1.044388e-03
## sample_typeNasal                       0.1552804    25464.53  6.097908e-06
## sample_typeSputum                     -8.4802015  1614863.76 -5.251342e-06
## treatmentlyPMA                        20.0406740    20795.51  9.637019e-04
## treatmentBenzonase                    -3.8606553   144818.73 -2.665854e-05
## treatmentHostZERO                     20.0406709    20795.51  9.637017e-04
## treatmentMolYsis                      20.0406701    20795.51  9.637017e-04
## treatmentQIAamp                       -5.4238408   313832.06 -1.728262e-05
## sample_typeNasal:treatmentlyPMA        3.2667884    25464.53  1.282878e-04
## sample_typeSputum:treatmentlyPMA     -22.3386079  5344185.65 -4.179984e-06
## sample_typeNasal:treatmentBenzonase    0.8974271   168154.90  5.336907e-06
## sample_typeSputum:treatmentBenzonase  -1.4086264 22564912.93 -6.242552e-08
## sample_typeNasal:treatmentHostZERO     0.8448602    25464.53  3.317792e-05
## sample_typeSputum:treatmentHostZERO  -20.3882571  2509755.81 -8.123602e-06
## sample_typeNasal:treatmentMolYsis      3.1358205    25464.53  1.231446e-04
## sample_typeSputum:treatmentMolYsis   -19.6253680  2080626.07 -9.432434e-06
## sample_typeNasal:treatmentQIAamp      -3.2248662  1761497.88 -1.830752e-06
## sample_typeSputum:treatmentQIAamp     -0.4359967 30057041.91 -1.450564e-08
##                                       Pr(>|z|)
## (Intercept)                          0.9991667
## sample_typeNasal                     0.9999951
## sample_typeSputum                    0.9999958
## treatmentlyPMA                       0.9992311
## treatmentBenzonase                   0.9999787
## treatmentHostZERO                    0.9992311
## treatmentMolYsis                     0.9992311
## treatmentQIAamp                      0.9999862
## sample_typeNasal:treatmentlyPMA      0.9998976
## sample_typeSputum:treatmentlyPMA     0.9999967
## sample_typeNasal:treatmentBenzonase  0.9999957
## sample_typeSputum:treatmentBenzonase 1.0000000
## sample_typeNasal:treatmentHostZERO   0.9999735
## sample_typeSputum:treatmentHostZERO  0.9999935
## sample_typeNasal:treatmentMolYsis    0.9999017
## sample_typeSputum:treatmentMolYsis   0.9999925
## sample_typeNasal:treatmentQIAamp     0.9999985
## sample_typeSputum:treatmentQIAamp    1.0000000

None of associations were significant at binomial generalized mixed effect model.

With glmer(lib_failed ~ sample_type + (1|subject_id)),

glmer(lib_failed ~ sample_type + (1|subject_id),
      data = sample_data %>% 
              subset(sample_type %in% c("BAL", "Nasal", "Sputum")) %>% 
              data.frame(check.names = F),
      family = binomial) %>%
        summary

## Generalized linear mixed model fit by maximum likelihood (Laplace
##   Approximation) [glmerMod]
##  Family: binomial  ( logit )
## Formula: lib_failed ~ sample_type + (1 | subject_id)
##    Data: 
## sample_data %>% subset(sample_type %in% c("BAL", "Nasal", "Sputum")) %>%  
##     data.frame(check.names = F)
## 
##      AIC      BIC   logLik deviance df.resid 
##     69.4     79.6    -30.7     61.4       91 
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -0.6325 -0.6325 -0.3333  0.0000  3.0000 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0        0       
## Number of obs: 95, groups:  subject_id, 20
## 
## Fixed effects:
##                     Estimate Std. Error z value Pr(>|z|)    
## (Intercept)       -2.197e+00  6.086e-01  -3.610 0.000306 ***
## sample_typeNasal   1.281e+00  7.144e-01   1.793 0.072970 .  
## sample_typeSputum -3.377e+01  1.179e+07   0.000 0.999998    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) smpl_N
## smpl_typNsl -0.852       
## smpl_typSpt  0.000  0.000
## optimizer (Nelder_Mead) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')

It seems like there were sample_type specific effects (since some of the sample * treatment did not shoy any failure).

Library failure (stratified)

glm ( sequencing fail ~ treatment + subject_id )

–> Cannot run for Sputum (no failed sample).

For BAL

GLMER result

glmer_libfail_bal <- glmer(lib_failed ~ treatment + (1|subject_id), data = sample_data %>% data.frame %>% dplyr::filter(sample_type %in% c("BAL")), family = "binomial")
glmer_libfail_bal %>% summary

## Generalized linear mixed model fit by maximum likelihood (Laplace
##   Approximation) [glmerMod]
##  Family: binomial  ( logit )
## Formula: lib_failed ~ treatment + (1 | subject_id)
##    Data: sample_data %>% data.frame %>% dplyr::filter(sample_type %in%  
##     c("BAL"))
## 
##      AIC      BIC   logLik deviance df.resid 
##     28.7     38.5     -7.3     14.7       23 
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -0.7529 -0.3404  0.0000  0.0000  1.9092 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 1.294    1.138   
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error z value Pr(>|z|)
## (Intercept)          -26.24    1430.39  -0.018    0.985
## treatmentlyPMA        24.49    1430.39   0.017    0.986
## treatmentBenzonase   -34.45    3045.52  -0.011    0.991
## treatmentHostZERO     24.49    1430.39   0.017    0.986
## treatmentMolYsis      24.49    1430.39   0.017    0.986
## treatmentQIAamp     -156.35    3045.52  -0.051    0.959
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -1.000                            
## trtmntBnzns  0.188 -0.188                     
## trtmntHZERO -1.000  1.000 -0.188              
## trtmntMlYss -1.000  1.000 -0.188  1.000       
## trtmntQIAmp  0.188 -0.188 -0.206 -0.188 -0.188
## optimizer (Nelder_Mead) convergence code: 0 (OK)
##  Hessian is numerically singular: parameters are not uniquely determined

No treatment significantly affected for library failure for BAL.

For nasal, GLMER result does not converge. As all the variables are categorical, we cannot rescale.

Host ratio (all samples)

None-stratified. p-value of interaction term was 4.587e-16

lmer_host_ratio_all <- lmer(sequencing_host_prop * 100 ~ sample_type * treatment + (1|subject_id), 
                    data = sample_data %>%
                            data.frame %>%
                            mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%
                            otu_table %>%
                            colSums()) %>%
                            subset(., total_read != 0))
        
lmer_host_ratio_all

## Linear mixed model fit by REML ['lmerModLmerTest']
## Formula: 
## sequencing_host_prop * 100 ~ sample_type * treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%  
##     otu_table %>% colSums()) %>% subset(., total_read != 0)
## REML criterion at convergence: 1052.31
## Random effects:
##  Groups     Name        Std.Dev.
##  subject_id (Intercept) 11.44   
##  Residual               12.04   
## Number of obs: 155, groups:  subject_id, 22
## Fixed Effects:
##                          (Intercept)                       sample_typeMock  
##                              15.8049                              -15.5516  
##                       sample_typeBAL                      sample_typeNasal  
##                              85.3037                               79.4212  
##                    sample_typeSputum                        treatmentlyPMA  
##                              83.1787                               -1.2538  
##                   treatmentBenzonase                     treatmentHostZERO  
##                               8.2009                                5.1396  
##                     treatmentMolYsis                       treatmentQIAamp  
##                               3.4039                                1.2271  
##       sample_typeMock:treatmentlyPMA         sample_typeBAL:treatmentlyPMA  
##                               6.6544                               -2.6414  
##      sample_typeNasal:treatmentlyPMA      sample_typeSputum:treatmentlyPMA  
##                             -26.5524                               -2.5511  
##   sample_typeMock:treatmentBenzonase     sample_typeBAL:treatmentBenzonase  
##                              -7.8214                              -10.8156  
##  sample_typeNasal:treatmentBenzonase  sample_typeSputum:treatmentBenzonase  
##                             -28.4482                              -14.4549  
##    sample_typeMock:treatmentHostZERO      sample_typeBAL:treatmentHostZERO  
##                              -4.6623                              -24.8644  
##   sample_typeNasal:treatmentHostZERO   sample_typeSputum:treatmentHostZERO  
##                             -78.9810                              -50.6196  
##     sample_typeMock:treatmentMolYsis       sample_typeBAL:treatmentMolYsis  
##                              -3.0151                              -22.5705  
##    sample_typeNasal:treatmentMolYsis    sample_typeSputum:treatmentMolYsis  
##                             -53.8806                              -73.0365  
##      sample_typeMock:treatmentQIAamp        sample_typeBAL:treatmentQIAamp  
##                              -0.8505                               -8.9627  
##     sample_typeNasal:treatmentQIAamp     sample_typeSputum:treatmentQIAamp  
##                             -76.3805                              -19.9221

ANOVA result

lmer_host_ratio_all %>% anova() 

Table S1. Host ratio (stratified)

Table S1. Effect size, standard error (SE) and p-value at a statistical test on host DNA ratio using linear mixed effect model with treatment as fixed effect and subject as random effect using r package lmer::lmer( Host DNA % ~ treatment + (1|subject_id) ). Stratified analyses were conducted for each sample type as an interaction term of sample type and treatment was significant at an ANOVA test (p-value < 0.001) using a model, LMER (Host DNA % ~ sample type + treatment + sample type * treatment + (1|subject_id) ). The baseline of categorical variable is untreated, and statistical significances were noted with *: p-value < 0.05 and ***: p-value < 0.001.

HostZERO and Molysius was effect to to all QIAamp was effective for Nasal swab and Sputum

Stratified (BAL)

hr_lmer_bal <- lmer(sequencing_host_prop * 100 ~ treatment + (1|subject_id), 
                    data = sample_data %>%
                            data.frame %>%
                            mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%
                            otu_table %>%
                            colSums()) %>%
                            subset(., .$sample_type %in% c("BAL")))

hr_lmer_bal %>%
        summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: sequencing_host_prop * 100 ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%  
##     otu_table %>% colSums()) %>% subset(., .$sample_type %in%      c("BAL"))
## 
## REML criterion at convergence: 196.7
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -3.4797 -0.1994 -0.0472  0.5522  1.1072 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept)  35.98    5.998  
##  Residual               119.77   10.944  
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          99.639      5.581  18.946  17.853 2.62e-13 ***
## treatmentlyPMA       -3.123      6.922  20.000  -0.451   0.6567    
## treatmentBenzonase   -1.145      6.922  20.000  -0.165   0.8703    
## treatmentHostZERO   -18.255      6.922  20.000  -2.637   0.0158 *  
## treatmentMolYsis    -17.697      6.922  20.000  -2.557   0.0188 *  
## treatmentQIAamp      -6.266      6.922  20.000  -0.905   0.3761    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.620                            
## trtmntBnzns -0.620  0.500                     
## trtmntHZERO -0.620  0.500  0.500              
## trtmntMlYss -0.620  0.500  0.500  0.500       
## trtmntQIAmp -0.620  0.500  0.500  0.500  0.500

Stratified (nasal)

hr_lmer_ns <- lmer(sequencing_host_prop * 100 ~ treatment + (1|subject_id), data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal")))

hr_lmer_ns %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: sequencing_host_prop * 100 ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal"))
## 
## REML criterion at convergence: 275.3
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.6062 -0.5120 -0.1037  0.6480  1.5793 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 179.0    13.38   
##  Residual               417.5    20.43   
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          95.226      7.723  24.557  12.330 5.16e-12 ***
## treatmentlyPMA      -27.681     11.631  23.874  -2.380 0.025655 *  
## treatmentBenzonase  -19.986     11.657  24.145  -1.715 0.099245 .  
## treatmentHostZERO   -73.580     11.657  24.145  -6.312 1.55e-06 ***
## treatmentMolYsis    -50.602     11.631  23.874  -4.351 0.000219 ***
## treatmentQIAamp     -75.415     11.657  24.145  -6.470 1.06e-06 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.465                            
## trtmntBnzns -0.464  0.255                     
## trtmntHZERO -0.464  0.255  0.385              
## trtmntMlYss -0.465  0.235  0.361  0.361       
## trtmntQIAmp -0.464  0.361  0.229  0.229  0.255

Stratified (sputum)

hr_lmer_spt <- lmer(sequencing_host_prop * 100 ~ treatment + (1|subject_id), data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum")))

hr_lmer_spt %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: sequencing_host_prop * 100 ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum"))
## 
## REML criterion at convergence: 192.4
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -2.25995 -0.43375  0.06531  0.47119  1.49643 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept)  25.39    5.039  
##  Residual               101.92   10.096  
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          98.984      5.046  20.018  19.616 1.53e-14 ***
## treatmentlyPMA       -3.805      6.385  20.000  -0.596  0.55792    
## treatmentBenzonase   -6.254      6.385  20.000  -0.979  0.33904    
## treatmentHostZERO   -45.480      6.385  20.000  -7.123 6.68e-07 ***
## treatmentMolYsis    -69.633      6.385  20.000 -10.906 7.22e-10 ***
## treatmentQIAamp     -18.695      6.385  20.000  -2.928  0.00832 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.633                            
## trtmntBnzns -0.633  0.500                     
## trtmntHZERO -0.633  0.500  0.500              
## trtmntMlYss -0.633  0.500  0.500  0.500       
## trtmntQIAmp -0.633  0.500  0.500  0.500  0.500

Merged table

hr_lmer_bal_kbl <- hr_lmer_bal %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              hr_lmer_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]

hr_lmer_ns_kbl <- hr_lmer_ns %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              hr_lmer_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        



hr_lmer_spt_kbl <- hr_lmer_spt %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              hr_lmer_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        


tables1 <- cbind(hr_lmer_bal_kbl, hr_lmer_ns_kbl, hr_lmer_spt_kbl) %>%
    kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 1, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")
tables1

save_kable(tables1, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS1.html", self_contained = T)

Final reads (all samples)

None-stratified. p-value of interaction term was 3.955e-09

Final reads output

lmer_final_reads <- lmer(Final_reads * 100 ~ sample_type * treatment + (1|subject_id), 
                    data = sample_data %>%
                            data.frame %>%
                            mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%
                            otu_table %>%
                            colSums()) %>%
                            subset(., total_read != 0))

lmer_final_reads %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: Final_reads * 100 ~ sample_type * treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%  
##     otu_table %>% colSums()) %>% subset(., total_read != 0)
## 
## REML criterion at convergence: 5743.8
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -2.2476 -0.3384 -0.0107  0.1503  4.4188 
## 
## Random effects:
##  Groups     Name        Variance  Std.Dev. 
##  subject_id (Intercept) 5.197e+16 2.280e+08
##  Residual               3.548e+18 1.884e+09
## Number of obs: 155, groups:  subject_id, 22
## 
## Fixed effects:
##                                        Estimate Std. Error         df t value
## (Intercept)                           1.185e+08  8.021e+08  1.064e+01   0.148
## sample_typeMock                       9.198e+09  1.134e+09  1.064e+01   8.109
## sample_typeBAL                       -8.967e+07  1.242e+09  3.953e+01  -0.072
## sample_typeNasal                      4.114e+08  1.002e+09  2.139e+01   0.411
## sample_typeSputum                    -4.727e+07  1.168e+09  3.319e+01  -0.040
## treatmentlyPMA                        1.957e+09  1.141e+09  3.622e+19   1.716
## treatmentBenzonase                   -2.317e+07  1.141e+09  2.654e+18  -0.020
## treatmentHostZERO                     3.892e+07  1.141e+09  8.054e+18   0.034
## treatmentMolYsis                     -6.575e+06  1.141e+09  3.917e+18  -0.006
## treatmentQIAamp                      -2.342e+06  1.141e+09  1.867e+18  -0.002
## sample_typeMock:treatmentlyPMA       -6.842e+09  1.613e+09  2.346e+19  -4.241
## sample_typeBAL:treatmentlyPMA        -1.885e+09  1.754e+09  1.069e+20  -1.075
## sample_typeNasal:treatmentlyPMA      -2.353e+09  1.540e+09  1.617e+04  -1.528
## sample_typeSputum:treatmentlyPMA     -1.744e+09  1.649e+09  4.737e+19  -1.058
## sample_typeMock:treatmentBenzonase   -2.499e+08  1.613e+09  8.166e+18  -0.155
## sample_typeBAL:treatmentBenzonase     2.079e+08  1.703e+09  1.086e+05   0.122
## sample_typeNasal:treatmentBenzonase   4.019e+08  1.540e+09  1.588e+04   0.261
## sample_typeSputum:treatmentBenzonase  4.401e+08  1.649e+09  1.045e+19   0.267
## sample_typeMock:treatmentHostZERO     5.152e+08  1.613e+09  1.197e+19   0.319
## sample_typeBAL:treatmentHostZERO      5.142e+08  1.703e+09  1.086e+05   0.302
## sample_typeNasal:treatmentHostZERO    3.692e+09  1.540e+09  1.588e+04   2.398
## sample_typeSputum:treatmentHostZERO   3.483e+09  1.649e+09  1.874e+19   2.112
## sample_typeMock:treatmentMolYsis     -4.309e+08  1.613e+09  8.904e+18  -0.267
## sample_typeBAL:treatmentMolYsis       8.874e+08  1.703e+09  1.086e+05   0.521
## sample_typeNasal:treatmentMolYsis     8.013e+08  1.540e+09  1.617e+04   0.520
## sample_typeSputum:treatmentMolYsis    6.766e+09  1.649e+09  3.134e+19   4.103
## sample_typeMock:treatmentQIAamp       2.329e+09  1.613e+09  7.701e+18   1.444
## sample_typeBAL:treatmentQIAamp        8.553e+08  1.703e+09  1.086e+05   0.502
## sample_typeNasal:treatmentQIAamp      3.664e+09  1.540e+09  1.588e+04   2.380
## sample_typeSputum:treatmentQIAamp     2.157e+09  1.649e+09  4.723e+18   1.308
##                                      Pr(>|t|)    
## (Intercept)                            0.8853    
## sample_typeMock                      7.08e-06 ***
## sample_typeBAL                         0.9428    
## sample_typeNasal                       0.6853    
## sample_typeSputum                      0.9680    
## treatmentlyPMA                         0.0862 .  
## treatmentBenzonase                     0.9838    
## treatmentHostZERO                      0.9728    
## treatmentMolYsis                       0.9954    
## treatmentQIAamp                        0.9984    
## sample_typeMock:treatmentlyPMA       2.22e-05 ***
## sample_typeBAL:treatmentlyPMA          0.2825    
## sample_typeNasal:treatmentlyPMA        0.1264    
## sample_typeSputum:treatmentlyPMA       0.2902    
## sample_typeMock:treatmentBenzonase     0.8769    
## sample_typeBAL:treatmentBenzonase      0.9028    
## sample_typeNasal:treatmentBenzonase    0.7941    
## sample_typeSputum:treatmentBenzonase   0.7896    
## sample_typeMock:treatmentHostZERO      0.7494    
## sample_typeBAL:treatmentHostZERO       0.7627    
## sample_typeNasal:treatmentHostZERO     0.0165 *  
## sample_typeSputum:treatmentHostZERO    0.0347 *  
## sample_typeMock:treatmentMolYsis       0.7893    
## sample_typeBAL:treatmentMolYsis        0.6023    
## sample_typeNasal:treatmentMolYsis      0.6027    
## sample_typeSputum:treatmentMolYsis   4.08e-05 ***
## sample_typeMock:treatmentQIAamp        0.1488    
## sample_typeBAL:treatmentQIAamp         0.6155    
## sample_typeNasal:treatmentQIAamp       0.0173 *  
## sample_typeSputum:treatmentQIAamp      0.1908    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

ANOVA on lmer result

lmer_final_reads %>% anova() 

Table S2 Final reads (stratified)

lmer( Host DNA ratio ~ treatment + (1|subject_id) )

Table S2. Changes on final reads stratified by sample type tested with linear mixed effect models using r package lmer::lmer(log10(Final reads) ~ Treatment + (1|Subject id)). Stratified analyses were conducted for each sample type as an interaction term of sample type and treatment was significant at an ANOVA test (p-value < 0.001) using a model, LMER (log10(Final reads) ~ sample type + treatment + sample type * treatment + (1|subject_id) ). Effect size with adjusted 95% confidence intervals and p-value were listed. The unit of final read is reads x 106. Statistical significances were noted with *: p-value < 0.05, **: p-value < 0.01 and ***: p-value < 0.001.

Sputum’s final read increased after every treatment. Nasal swab showed improved reads with lyPMA, HostZERO and QIAamp. BAL also showed increased reads with most of treatment.

Raw results, BAL, nasal and Sputum, respectively

fr_lmer_bal <- lmer(log10(Final_reads/1000000) ~ treatment + (1|subject_id),
                    data = sample_data %>%
                                data.frame %>%
                                mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%
                                otu_table %>%
                                colSums()) %>%
             subset(., .$sample_type %in% c("BAL")))


fr_lmer_bal %>% 
        summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log10(Final_reads/1e+06) ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%  
##     otu_table %>% colSums()) %>% subset(., .$sample_type %in%      c("BAL"))
## 
## REML criterion at convergence: 45.6
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.65203 -0.51284  0.03098  0.64282  1.40913 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.07389  0.2718  
##  Residual               0.21756  0.4664  
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)   
## (Intercept)         -0.5067     0.2414 18.1625  -2.099  0.05009 . 
## treatmentlyPMA       0.3521     0.2950 20.0000   1.194  0.24656   
## treatmentBenzonase   0.8109     0.2950 20.0000   2.749  0.01238 * 
## treatmentHostZERO    0.9502     0.2950 20.0000   3.221  0.00429 **
## treatmentMolYsis     1.0358     0.2950 20.0000   3.511  0.00220 **
## treatmentQIAamp      1.0481     0.2950 20.0000   3.553  0.00199 **
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.611                            
## trtmntBnzns -0.611  0.500                     
## trtmntHZERO -0.611  0.500  0.500              
## trtmntMlYss -0.611  0.500  0.500  0.500       
## trtmntQIAmp -0.611  0.500  0.500  0.500  0.500

fr_lmer_ns <- lmer(log10(Final_reads/1000000) ~ treatment + (1|subject_id),
                    data = sample_data %>%
                                data.frame %>%
                                mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%
                                otu_table %>%
                                colSums()) %>%
             subset(., .$sample_type %in% c("Nasal")))


fr_lmer_ns %>% 
        summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log10(Final_reads/1e+06) ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%  
##     otu_table %>% colSums()) %>% subset(., .$sample_type %in%      c("Nasal"))
## 
## REML criterion at convergence: 52.3
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.0897 -0.7843 -0.2293  0.5901  1.6955 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.07592  0.2755  
##  Residual               0.19354  0.4399  
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          0.5170     0.1642 25.0746   3.150 0.004193 ** 
## treatmentlyPMA      -0.5413     0.2499 24.0343  -2.166 0.040468 *  
## treatmentBenzonase   0.1437     0.2505 24.2989   0.574 0.571339    
## treatmentHostZERO    0.8560     0.2505 24.2989   3.418 0.002229 ** 
## treatmentMolYsis     0.2107     0.2499 24.0343   0.843 0.407498    
## treatmentQIAamp      1.0502     0.2505 24.2989   4.193 0.000316 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.472                            
## trtmntBnzns -0.471  0.259                     
## trtmntHZERO -0.471  0.259  0.383              
## trtmntMlYss -0.472  0.239  0.359  0.359       
## trtmntQIAmp -0.471  0.359  0.234  0.234  0.259

fr_lmer_spt <- lmer(log10(Final_reads/1000000) ~ treatment + (1|subject_id),
                    data = sample_data %>%
                                data.frame %>%
                                mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%
                                otu_table %>%
                                colSums()) %>%
             subset(., .$sample_type %in% c("Sputum"))) 


fr_lmer_spt %>% 
        summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: log10(Final_reads/1e+06) ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% mutate(total_read = phyloseq_unfiltered$phyloseq_count %>%  
##     otu_table %>% colSums()) %>% subset(., .$sample_type %in%      c("Sputum"))
## 
## REML criterion at convergence: 8.4
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.89244 -0.59432 -0.06631  0.55937  1.70987 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.00774  0.08798 
##  Residual               0.04975  0.22305 
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)         -0.1734     0.1072 22.0056  -1.617  0.12010    
## treatmentlyPMA       0.5401     0.1411 20.0000   3.828  0.00105 ** 
## treatmentBenzonase   0.8459     0.1411 20.0000   5.996 7.30e-06 ***
## treatmentHostZERO    1.6733     0.1411 20.0000  11.862 1.67e-10 ***
## treatmentMolYsis     1.9901     0.1411 20.0000  14.107 7.42e-12 ***
## treatmentQIAamp      1.4158     0.1411 20.0000  10.036 2.98e-09 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.658                            
## trtmntBnzns -0.658  0.500                     
## trtmntHZERO -0.658  0.500  0.500              
## trtmntMlYss -0.658  0.500  0.500  0.500       
## trtmntQIAmp -0.658  0.500  0.500  0.500  0.500

Tidified output

fr_lmer_bal_kbl <- fr_lmer_bal %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              fr_lmer_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
fr_lmer_ns_kbl <- fr_lmer_ns %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              fr_lmer_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


fr_lmer_spt_kbl <- fr_lmer_spt %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              fr_lmer_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


tables2 <- cbind(fr_lmer_bal_kbl, fr_lmer_ns_kbl, fr_lmer_spt_kbl) %>%
    kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 1, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")
tables2

save_kable(tables2, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS2.html", self_contained = T)

Fig. S3. Figure of sequencing result

ii. How were changes in sequencing results?

Fig. S3. Host depletion effects measured by shotgun metagenomic sequencing. (A) Raw DNA reads, (B) host mapped reads by bowtie2, (C) final reads of microbes, and (D) proportion of host mapped among total mapped reads.

fs3a <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = log10(Raw_reads))) +
        geom_jitter(aes(col = treatment, x = treatment),
                    lwd = 0.2, alpha = 0.3, stroke = 0)+
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        ylab("log<sub>10</sub>(raw reads)") +
        labs(tag = "A") +
        facet_wrap(~sample_type, scale = "free_x") +
        guides(fill = guide_legend(nrow = 1))

#   - Host_mapped


fs3b <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = log10(Host_mapped))) +
        geom_jitter(aes(col = treatment, x = treatment),
                    lwd = 0.2, alpha = 0.3, stroke = 0)+
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        ylab("log<sub>10</sub>(host reads)") +
        labs(tag = "B") +
        facet_wrap(~sample_type, scale = "free_x") +
        guides(fill = guide_legend(nrow = 1))







#   - % Host (we have used Host_mapped/Raw_reads in prior papers)

#   - Final_reads

fs3c <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = log10(Final_reads))) +
        geom_jitter(aes(col = treatment, x = treatment),
                    lwd = 0.2, alpha = 0.3, stroke = 0)+
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        ylab("log<sub>10</sub>(final reads)") +
        labs(tag = "C") +
        facet_wrap(~sample_type, scale = "free_x") +
        guides(fill = guide_legend(nrow = 1))





#   - % Host (we have used Host_mapped/Raw_reads in prior papers)
fs3d <- ggplot(sample_data %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")), aes(x = sample_type, y = sequencing_host_prop)) +
        geom_jitter(aes(col = treatment, x = treatment),
                    lwd = 0.2, alpha = 0.3, stroke = 0)+
        #geom_linerange(aes(xmin=lower.ci, xmax=upper.ci)) +
        stat_summary(aes(color = treatment, x = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        theme_classic (base_size = 12, base_family = "sans")+ 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        scale_x_discrete(name ="Treatment")+
        theme(axis.title.y = element_markdown(),
              plot.tag = element_text(size = 15),
              axis.text.x = element_blank()) +
        
        ylab("Host ratio by sequencing") +
        labs(tag = "D") +
        facet_wrap(~sample_type, scale = "free_x") +
        guides(fill = guide_legend(nrow = 1))



figS3 <- ggarrange(fs3a, fs3b, fs3c, fs3d, common.legend = T, align = "hv")

figS3

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS3.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 170, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

figS3
# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Table S10. decontam - stratified by sample_type

Table S10. Summary table of potential contaminants and their prevalence across all samples (N = 113) and within negative controls (total N = 31). The contaminants were identified using decontam37 combined method (Fisher’s exact test result of prevalence and frequency method results), using 16S qPCR bacterial DNA concentration as total bacterial load at prevalence threshold of 0.1. The analyses were stratified by sample types.

#Stratified by sample type
neg_number_decontam <- 
        phyloseq_unfiltered$phyloseq_count %>% 
        subset_samples(sample_type == "Neg.") %>%
        subset_samples(S.obs != 0) %>%
        sample_names() %>%
        length

sample_number_decontam <- 
        phyloseq_unfiltered$phyloseq_count %>% 
        subset_samples(sample_type != "Mock") %>%
        subset_samples(sample_type != "Neg.") %>%
        subset_samples(S.obs != 0) %>%
        sample_names() %>%
        length

prev_neg <- ((phyloseq_unfiltered$phyloseq_count %>% 
          subset_samples(sample_type == "Neg.") %>%
          otu_table %>% 
          data.frame(.)) != 0) %>%
        rowSums %>% data.frame %>% rename("Prevalence (negative controls)" = ".")

prev_all <- ((phyloseq_unfiltered$phyloseq_count %>%
                      subset_samples(sample_type != "Mock") %>%
                      subset_samples(sample_type != "Neg.") %>%
          otu_table %>% 
          data.frame(.)) != 0) %>%
        rowSums %>% data.frame %>% rename("Prevalence (all)" = ".")


sample_data(phyloseq_unfiltered$phyloseq_rel)$is.neg <- grepl("Neg", sample_data(phyloseq_unfiltered$phyloseq_rel)$sample_type)

phyloseq_decontam_bal <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(sample_type == "Neg." | sample_type == "BAL")
phyloseq_decontam_ns <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(sample_type == "Neg." | sample_type == "Nasal")
phyloseq_decontam_spt <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(sample_type == "Neg." | sample_type == "Sputum")


contaminant_combined_bal <- 
data.frame("BAL", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_bal, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)

contaminant_combined_ns <- 
data.frame("Nasal swab", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_ns, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)


contaminant_combined_spt <- 
data.frame("Sputum", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_spt, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)

contaminants <- rbind(contaminant_combined_bal, contaminant_combined_ns, contaminant_combined_spt)

names(contaminants) <- c("Sample type", "Taxa")


neg_number_decontam

## [1] 31

sample_number_decontam

## [1] 93

merged_contaminants <- merge(contaminants, prev_all %>% 
                                     rownames_to_column("Taxa"), by = "Taxa") %>%
        merge(., prev_neg %>%
                      rownames_to_column("Taxa"), by = "Taxa") %>%
       dplyr::select(c("Sample type",
                       "Taxa",
                       "Prevalence (all)",
                       "Prevalence (negative controls)")) %>%
        mutate(`Prevalence (all)` = paste0(`Prevalence (all)`,
                                           " (",
                                           round(`Prevalence (all)`/
                                                         sample_number_decontam * 100, 2),
                                           "%)"),
                `Prevalence (negative controls)` = paste0(`Prevalence (negative controls)`,
                                           " (",
                                           round(`Prevalence (negative controls)`/
                                                         neg_number_decontam * 100, 2),
                                           "%)")) %>%
        rename(`Prevalence (all) N = 93` = `Prevalence (all)`,
               `Prevalence (negative controls) N = 31` = `Prevalence (negative controls)`) %>%
        .[order(.$"Sample type", .$"Taxa"),] %>%
        remove_rownames()


species_italic2 <- function(data){
  data <- gsub("_", " ", data)
  data <- gsub("[]]|[[]", "", data)
  data <- gsub(" sp", " sp.", data)
  data <- gsub(" sp.", "</em> sp.", data)
  data <- gsub(" group", "", data)
  data <- ifelse(grepl("[*]", data), paste("<em>", data, sep = ""), paste("<em>", data, "</em>", sep = ""))
  data
}

Decontam result

Stratified-BAL

isContaminant(phyloseq_decontam_bal, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) 

Stratified-Nasal

isContaminant(phyloseq_decontam_ns, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) 

Stratified-Sputum

isContaminant(phyloseq_decontam_spt, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) 

Decontam -stratified by treatment

#Stratified by sample type

prev_neg <- ((phyloseq_unfiltered$phyloseq_count %>% 
          subset_samples(sample_type == "Neg.") %>%
          otu_table %>% 
          data.frame(.)) != 0) %>%
        rowSums %>% data.frame %>% rename("Prevalence (negative controls)" = ".")

prev_all <- ((phyloseq_unfiltered$phyloseq_count %>%
                      subset_samples(sample_type != "Mock") %>%
                      subset_samples(sample_type != "Neg.") %>%
          otu_table %>% 
          data.frame(.)) != 0) %>%
        rowSums %>% data.frame %>% rename("Prevalence (all)" = ".")


sample_data(phyloseq_unfiltered$phyloseq_rel)$is.neg <- grepl("Neg", sample_data(phyloseq_unfiltered$phyloseq_rel)$sample_type)

phyloseq_decontam_lypma <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(treatment == "Untreated" | treatment == "lyPMA")

phyloseq_decontam_ben <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(treatment == "Untreated" | treatment == "Benzonase")

phyloseq_decontam_hostzero <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(treatment == "Untreated" | treatment == "HostZero")

phyloseq_decontam_molysis <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(treatment == "Untreated" | treatment == "MolYsis")

phyloseq_decontam_qiaamp <- phyloseq_unfiltered$phyloseq_rel %>% 
        subset_samples(S.obs != 0) %>%
        subset_samples(treatment == "Untreated" | treatment == "QIAamp")


contaminant_combined_lypma <- 
data.frame("lyPMA", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_lypma, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)

contaminant_combined_ben <- 
data.frame("Benzonase", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_ben, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)


contaminant_combined_hostzero <- 
data.frame("HostZero", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_hostzero, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)


contaminant_combined_molysis <- 
data.frame("MolYsis", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_molysis, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)


contaminant_combined_QIAamp <- 
data.frame("QIAamp", fix.empty.names = F, 
        isContaminant(phyloseq_decontam_qiaamp, method="combined", neg = "is.neg", threshold = 0.1, conc = "DNA_bac_ng_uL") %>% subset(.,.$contaminant) %>% row.names
)

contaminants_strat_treat <- rbind(contaminant_combined_lypma,
                      contaminant_combined_ben, 
                      contaminant_combined_hostzero,
                      contaminant_combined_molysis,
                      contaminant_combined_QIAamp
                      )

names(contaminants_strat_treat) <- c("Treatment", "Taxa")

12 taxa identified as contaminants. 7 Taxa identified as contaminants (sample type stratified).

Decontam summary

Summary table of potential contaminants with all sample types and stratified sample types in all methods (prevalence, frequence, and combined)

tableS10 <- merged_contaminants %>% 
        mutate(Taxa = species_italic2(Taxa)
               ) %>%
        kbl(format = "html", escape = F) %>%
        kable_styling(full_width = 0, html_font = "sans") 
tableS10

save_kable(tableS10, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS10.html", self_contained = T)

Prevalence filtering

Prevalence & abundance filtering was conducted

22. M&M - Prevalence filtering

    1.  Prevalence filtration at 5%, except its abundance is over 0.75 quantile.
            a.  Information in the main text

phyloseq_unfiltered$phyloseq_rel <- transform_sample_counts(phyloseq_unfiltered$phyloseq_rel,
                                                            function(x){x/sum(x)})
taxa_qc <- data.frame("species" =
                              otu_table(
                                      subset_samples(
                                              phyloseq_unfiltered$phyloseq_rel,S.obs != 0 &
                                                      sample_type %in% c("Mock", "BAL", "Nasal", "Sputum"))) %>%
                              t() %>% colnames(),
                      "prevalence" =
                              ifelse(subset_samples(phyloseq_unfiltered$phyloseq_rel, S.obs != 0 & 
                                                            sample_type %in% c("Mock", "BAL", "Nasal", "Sputum")) %>%
                                             otu_table() > 0, 1, 0)%>% 
                              t() %>%
                              colSums(), #Prevalence of taxa
                      "mean_rel_abd" = 
                              subset_samples(phyloseq_unfiltered$phyloseq_rel,
                                             S.obs != 0 & sample_type %in% c("Mock", "BAL", "Nasal", "Sputum")) %>%
                              otu_table() %>%
                              t() %>%
                              colMeans(na.rm = T) #mean relativ abundacne 
)


function_qc <- data.frame("function" =
                                  otu_table(
                                          subset_samples(
                                                  phyloseq_unfiltered$phyloseq_path_rpk,
                                                           S.obs != 0 & 
                                                          sample_type %in% 
                                                          c("Mock", "BAL", "Nasal", "Sputum")
                                                  )
                                          ) %>% 
                                  t() %>% 
                                  colnames(),
                          "prevalence" = 
                                  ifelse(subset_samples(phyloseq_unfiltered$phyloseq_path_rpk,
                                                        S.obs != 0 & 
                                                                sample_type %in% 
                                                                c("Mock", "BAL", "Nasal", "Sputum")
                                                        ) %>% 
                                                 otu_table() > 0, 
                                         1, 
                                         0
                                         ) %>% 
                                  t() %>% 
                                  colSums(), #Prevalence of taxa
                          "mean_rpk" = 
                                  subset_samples(phyloseq_unfiltered$phyloseq_path_rpk, 
                                                 S.obs != 0 & 
                                                         sample_type %in% 
                                                         c("Mock", "BAL", "Nasal", "Sputum")
                                                 ) %>% 
                                  otu_table() %>% 
                                  t() %>% 
                                  colMeans(na.rm = T), #mean relativ abundacne 
                          unidentified = 
                                  ifelse((subset_samples(phyloseq_unfiltered$phyloseq_path_rpk,
                                              S.obs != 0 & sample_type %in% c("Mock", "BAL", "Nasal", "Sputum")) %>% 
                                       otu_table() > 0) %>%
                                      row.names() %in% c("UNMAPPED", "UNINTEGRATED")
                              , 1, 0)
)

red_flag_taxa <- data.frame(species = taxa_qc$species,
                            prevalence = taxa_qc$prevalence,
                            mean_rel_abd = taxa_qc$mean_rel_abd,
                            red_flag_prev_abd = 
                                    ifelse(taxa_qc$prevalence < 
                                                   otu_table(
                                                           subset_samples(
                                                                   phyloseq_unfiltered$phyloseq_rel,
                                                                   S.obs != 0 & 
                                                                           sample_type %in% 
                                                                           c("Mock", "BAL", "Nasal", "Sputum"))) %>%
                                                               t %>% rownames() %>%
                                                   length * 0.05 &
                                                   #Removing taxa with zero prevalence - taxa from nasal swabs
                                                   taxa_qc$mean_rel_abd <
                                                   taxa_qc %>%
                                                   subset(., .$prevalence != 0) %>%
                                                   .$mean_rel_abd %>%
                                                   quantile(., 0.75), 1,0),
                           red_flag_prev =
                                    ifelse(taxa_qc$prevalence < 
                                                   otu_table(
                                                           subset_samples(
                                                                   phyloseq_unfiltered$phyloseq_rel,
                                                                   S.obs != 0 & 
                                                                           sample_type %in% 
                                                                           c("Mock", "BAL", "Nasal", "Sputum"))) %>%
                                                               t %>% rownames() %>%
                                                   length * 0.05,
                                           1,
                                           0)) %>%
        mutate(red_flag_decontam = species %in% (contaminants$Taxa %>% unique()))

subset(red_flag_taxa, red_flag_taxa$red_flag_prev == 1 & red_flag_taxa$red_flag_prev_abd == 0)

#Unampped function were removed

red_flag_function <- 
        data.frame(function. = function_qc$function.,
                   prevalence = function_qc$prevalence,
                   mean_rel_abd = function_qc$mean_rpk,
                   red_flag_prev_abd = 
                           ifelse(function_qc$prevalence < 
                                          otu_table(
                                                  subset_samples(
                                                          phyloseq_unfiltered$phyloseq_path_rpk,
                                                          S.obs != 0 & 
                                                                  sample_type %in% 
                                                                  c("Mock", "BAL", "Nasal", "Sputum"))) %>%
                                          t %>% 
                                          rownames() %>%
                                          length * 0.05 &
                                          #Removing taxa with zero prevalence - taxa from nasal swabs
                                          function_qc$mean_rpk <
                                          function_qc %>%
                                          subset(., .$prevalence != 0) %>%
                                          .$mean_rpk %>%
                                          quantile(., 0.75), 1,0),
                   red_flag_prev =
                           ifelse(function_qc$prevalence <
                                          otu_table(
                                                  subset_samples(
                                                          phyloseq_unfiltered$phyloseq_path_rpk,
                                                          S.obs != 0 & 
                                                                  sample_type %in% 
                                                                  c("Mock", "BAL", "Nasal", "Sputum"))) %>%
                                          t %>% 
                                          rownames() %>%
                                          length * 0.05,
                                          1,
                                          0)) %>%
        mutate(red_flag_prev_abd = case_when(function. %in% c("UNMAPPED", "UNINTEGRATED") ~ 1,
                                     .default = red_flag_prev_abd))

subset(red_flag_function, red_flag_function$red_flag_prev == 1 & red_flag_function$red_flag_prev_abd == 0)

#decontaminated phyloseq 
phyloseq_decontam <- phyloseq
phyloseq_decontam$phyloseq_count <- prune_taxa(subset(red_flag_taxa,
                                           red_flag_taxa$red_flag_prev_abd != 1 &
                                                   !red_flag_taxa$red_flag_decontam)$species,
                                    phyloseq$phyloseq_count)

phyloseq_decontam$phyloseq_rel <- prune_taxa(subset(red_flag_taxa,
                                           red_flag_taxa$red_flag_prev_abd != 1 &
                                                   !red_flag_taxa$red_flag_decontam)$species,
                                    phyloseq$phyloseq_rel) %>%
        transform_sample_counts(., function(x){x/sum(x)})
        
#phyloseq for analysis
phyloseq$phyloseq_count <- prune_taxa(subset(red_flag_taxa,
                                           red_flag_taxa$red_flag_prev_abd != 1)$species,
                                    phyloseq$phyloseq_count)
phyloseq$phyloseq_rel <- prune_taxa(subset(red_flag_taxa,
                                           red_flag_taxa$red_flag_prev_abd != 1)$species,
                                    phyloseq$phyloseq_rel) %>%
        transform_sample_counts(., function(x){x/sum(x)})


phyloseq$phyloseq_path_rpk <- prune_taxa(subset(red_flag_function, red_flag_function$red_flag_prev_abd != 1)$function., phyloseq$phyloseq_path_rpk)

#phyloseq$tree_phyloseq_count <- prune_taxa(subset(red_flag_taxa,
                                           #red_flag_taxa$red_flag_prev_abd != 1 & !red_flag_taxa$red_flag_decontam_prev)$species,
                                    #phyloseq$tree_phyloseq_count)

#phyloseq$tree_phyloseq_rel <- prune_taxa(subset(red_flag_taxa,
                                           #red_flag_taxa$red_flag_prev_abd != 1 & !red_flag_taxa$red_flag_decontam_prev)$species,
                                    #phyloseq$tree_phyloseq_rel)

Filtration result

Taxa were filtered wehn red_flag_prev_abd == 1.

red_flag_taxa

Alpha diversity calculation

#Calculation of alpha diversity indices for filtered samples

alpha_diversity <- function(data) {
        otu_table <- otu_table(data) #%>% .[, colSums(.) !=0]
        S.obs <- rowSums(t(otu_table) != 0)
        sample_data <- sample_data(data)
        data_evenness <- vegan::diversity(t(otu_table)) / log(vegan::specnumber(t(otu_table))) # calculate evenness index using vegan package
        data_shannon <- vegan::diversity(t(otu_table), index = "shannon") # calculate Shannon index using vegan package
        data_hill <- exp(data_shannon)                           # calculate Hills index
        data_dominance <- microbiome::dominance(otu_table, index = "all", rank = 1, aggregate = TRUE) # dominance (Berger-Parker index), etc.
        data_invsimpson <- vegan::diversity(t(otu_table), index = "invsimpson")                          # calculate Shannon index using vegan package
        alpha_diversity <- cbind(S.obs, data_shannon, data_hill, data_invsimpson, data_evenness,data_dominance) # combine all indices in one data table
        sample_data <- merge(data.frame(sample_data), alpha_diversity, by = 0, all = T) %>% column_to_rownames(var = "Row.names")
}
#sample_data(phyloseq$phyloseq_count) <- sample_data(alpha_diversity(phyloseq$phyloseq_count))
sample_data(phyloseq$phyloseq_rel) <- sample_data(alpha_diversity(phyloseq$phyloseq_count))
sample_data(phyloseq$phyloseq_count) <-sample_data(alpha_diversity(phyloseq$phyloseq_count)) 
sample_data(phyloseq$phyloseq_path_rpk) <- sample_data(alpha_diversity(phyloseq$phyloseq_path_rpk))  

sample_data(phyloseq_tv) <- sample_data(alpha_diversity(phyloseq_tv))  

sample_data(phyloseq_decontam$phyloseq_rel) <- sample_data(alpha_diversity(phyloseq_decontam$phyloseq_count))  

sample_data(phyloseq_decontam$phyloseq_count) <- sample_data(alpha_diversity(phyloseq_decontam$phyloseq_count))

sample_data <- sample_data(phyloseq$phyloseq_count)

Alpha diversity calculation data

sample_data(phyloseq$phyloseq_rel) %>% 
        data.frame() %>%
        dplyr::select(c("S.obs", "data_shannon", "data_invsimpson")) 

Table 1 (revised)

Revised Table 1 was constructed after prevalence & abundacne filtering

Table 1. Sequencing results stratified by sample type and host depletion treatment. Number of samples in each experimental group, human and bacterial DNA measured by qPCR, number of samples did not pass the library QC or no microbial mapped reads, QC/d read, % host mapped reads, final reads, microbial species, function, and viral species richness. Microbial richness was calculated after employing prevalence and abundance filtration. Samples failed sequencing were removed from the calculation. Values depicted as N (%) or median (interquartile range).

sample_data(phyloseq$phyloseq_count)$sequencing_fail <- 
        ifelse(phyloseq$phyloseq_count %>% 
                       sample_data %>%
                       .$S.obs == 0,
               1,
               0)

sample_data(phyloseq$phyloseq_count) <- 
        
        merge(phyloseq$phyloseq_count %>% 
                      sample_data %>%
                      data.frame(check.names = F),
              phyloseq$phyloseq_path_rpk %>% #extracting functional richness as dataframe
                      sample_data %>%
                      data.frame(check.names = F) %>% 
                     dplyr::select(c("S.obs")) %>%
                      rename(F.obs = "S.obs"),
              by = 0
        ) %>%
                column_to_rownames("Row.names")

table1 <- sample_data(phyloseq$phyloseq_count) %>% data.frame() %>% 
        dplyr::filter(sample_type %in% c("Sputum", "Nasal", "BAL")) %>% 
        mutate(S.obs = case_when(S.obs==0 ~ NA,
                                 .default = S.obs),
               F.obs = case_when(F.obs==0 ~ NA,
                                 .default = F.obs),
               V.obs = case_when(V.obs==0 ~ NA,
                                 .default = V.obs)) %>% 
        group_by (sample_type, treatment) %>%
        summarise(`N` = n(),
            #      `Total DNA <br>ng/µL` = paste(format(round(median(picogreen_ng_ul),2), nsmall = 2, big.mark = ","), "<br>(", format(round(quantile(picogreen_ng_ul, 0.25),2), nsmall = 2, big.mark = ","), ", ", format(round(quantile(picogreen_ng_ul, 0.75),2), nsmall = 2, big.mark = ","), ")", sep = ""),
               `Human DNA <br>pg/µL` = paste(format(round(median(DNA_host_ng_uL*1000),1), nsmall = 1, big.mark = ","), " (", format(round(quantile(DNA_host_ng_uL*1000, 0.25),1), nsmall = 1, big.mark = ","), ", ", format(round(quantile(DNA_host_ng_uL*1000, 0.75),1), nsmall = 1, big.mark = ","), ")", sep = ""),
               `Bacterial DNA <br>pg/µL` = paste(format(round(median(DNA_bac_ng_uL*1000),1), nsmall = 1, big.mark = ","), " (", format(round(quantile(DNA_bac_ng_uL*1000, 0.25),1), nsmall = 1, big.mark = ","), ", ", format(round(quantile(DNA_bac_ng_uL*1000, 0.75),1), nsmall = 1, big.mark = ","), ")", sep = ""),
              `Sequencing fail<br>N (%)` = paste(sum(lib_failed + sequencing_fail), " (", sum(lib_failed +sequencing_fail) / n() * 100, " %)", sep = ""),
              `QC'd reads<br>reads x 10<sup>6</sup>` = paste(format(round(median(Reads_after_trim/1000000),1), nsmall = 1, big.mark = ","), " (", format(round(quantile(Reads_after_trim/1000000, 0.25),1), nsmall = 1, big.mark = ","), ", ", format(round(quantile(Reads_after_trim/1000000, 0.75),1), nsmall = 1, big.mark = ","), ")", sep = ""),
              `Host reads<br>%` = paste(format(round(median(sequencing_host_prop*100),1),
                                                nsmall = 1, big.mark = ","),
                                         " (",
                                         format(round(quantile(sequencing_host_prop * 100,
                                                               0.25),
                                                      1),
                                                nsmall = 1,
                                                big.mark = ","), 
                                         ", ", 
                                         format(round(quantile(sequencing_host_prop * 100, 0.75),1), 
                                                nsmall = 1, 
                                                big.mark = ","), 
                                         ")", 
                                         sep = ""),
             `Final reads<br>reads x 10<sup>6</sup>` = paste(format(round(median(Final_reads/1000000,
                                                                                 na.rm = T),
                                                                          1),
                                                                    nsmall = 1, big.mark = ","),
                                                             " (", 
                                                             format(round(quantile(Final_reads/1000000,
                                                                                            0.25,
                                                                                            na.rm = T),
                                                                                   1),
                                                                             nsmall = 1,
                                                                             big.mark = ","),
                                                             ", ",
                                                             format(round(quantile(Final_reads/1000000, 
                                                                                   0.75, 
                                                                                   na.rm = T),
                                                                          1), 
                                                                    nsmall = 1, 
                                                                    big.mark = ","), 
                                                             ")",
                                                             sep = ""),
             `Microbial<br>species<br>richness` = paste(median(S.obs, na.rm = T) %>% round,
                                                             " (", 
                                                             quantile(S.obs, 0.25, na.rm = T) %>% round,
                                                             ", ",
                                                             quantile(S.obs, 0.75, na.rm = T) %>% round,
                                                             ")",
                                                             sep = ""),
            `Microbial function<br>richness` = paste(median(F.obs, na.rm = T) %>% round,
                                                             " (", 
                                                             quantile(F.obs, 0.25, na.rm = T) %>% round,
                                                             ", ",
                                                             quantile(F.obs, 0.75, na.rm = T) %>% round,
                                                             ")",
                                                             sep = ""),
                `Viral<br>species<br>richness` = paste(median(V.obs, na.rm = T) %>% round,
                                                             " (", 
                                                             quantile(V.obs, 0.25, na.rm = T) %>% round,
                                                             ", ",
                                                             quantile(V.obs, 0.75, na.rm = T) %>% round,
                                                             ")",
                                                             sep = "")
              
            ) %>% 
        data.frame(check.names = F) %>% 
        arrange(sample_type, treatment) %>%
        rename(`Sample` = sample_type, Treatment = treatment) %>%
        mutate_all(linebreak) %>% kbl(format = "html", escape = F) %>% kable_styling(full_width = 0, html_font = "sans")

table1 

save_kable(table1, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/table1.html", self_contained = T)

mock_otu <- subset_samples(phyloseq$phyloseq_count, sample_type =="Mock") %>% otu_table() %>% data.frame()

mock_otu_nonzero <- mock_otu[rowSums(mock_otu[])>0,]

Figure 1 numbers (Host DNA) - treated vs untreated

This is calculations used for writing values of Figure 1 in the main text (as figure does not show exact value)

sample_data %>% 
        subset(., .$S.obs != 0) %>%
        group_by(sample_type, treated) %>%
        summarise(med = median(DNA_host_nondil + DNA_bac_nondil),
                  lo = quantile(DNA_host_ng_uL + DNA_host_ng_uL, 0.25),
                  high = quantile(DNA_host_nondil + DNA_bac_nondil, 0.75))

sample_data %>% subset(., .$S.obs != 0) %>%
        group_by(sample_type, treated) %>% summarise(med = median(DNA_host_ng_uL), lo = quantile(DNA_host_ng_uL, 0.25), high = quantile(DNA_host_ng_uL, 0.75))

Figure 1 numbers (Host prooportion) - treated vs untreated

sample_data %>% subset(., .$S.obs != 0) %>% group_by(sample_type, treated) %>% summarise(med = median(host_proportion), lo = quantile(host_proportion, 0.25), high = quantile(host_proportion, 0.75))

Figure 1 numbers (bacterial DNA) - treated vs untreated

sample_data %>% subset(., .$S.obs != 0) %>% group_by(sample_type, treated) %>% summarise(med = median(DNA_bac_ng_uL), lo = quantile(DNA_bac_ng_uL, 0.25), high = quantile(DNA_bac_ng_uL, 0.75))

Fig. S6. Bar plot of gram-stain

Fig. S6. Bar plot annotated with gram-stain information of (A) BAL, (B) nasal swabs, (C) sputum, and (D) mock communities after each treatment.

barplot_mock_gramproportion <- tax_table(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Mock")) %>%
        cbind(species20 = "[Other]") %>%
        {top20species <- head(taxa_sums(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Mock") %>% 
           subset_samples(., S.obs != 0)) %>%
                                data.frame %>%
                                arrange(-.) %>%
                                row.names(), 10)
   .[top20species, "species20"] <- as.character(.[top20species, "Species"])
   .[, 9] <- gsub("s__", " ", .[, 9])
   .[, 9] <- gsub("_", " ", .[, 9])
   .[, 9] <- gsub("[]]|[[]", "",  .[, 9])
   .[, 9] <- gsub(" sp", " sp.",  .[, 9])
   .[, 9] <- gsub(" sp.", "</i> sp.",  .[, 9])
   .[, 9] <- gsub(" group", "</i> group.",  .[, 9])
   .[, 9] <- ifelse(grepl("Other",.[, 9]),
                    "Other",
                    ifelse(grepl("</i>",  .[, 9]),
                           paste("<i>",  .[, 9], sep = ""),
                           paste("<i>",  .[, 9], "</i>", sep = "")) %>%
                            gsub("s__", "", .) %>%
                            gsub("_", " ", .)
   )
   phyloseq_temp <- subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Mock") %>% 
           subset_samples(., S.obs != 0)
   tax_table(phyloseq_temp) <- tax_table(.) 
   phyloseq_temp
  } %>%
        my_plot_bar(., fill="gram_stain") + 
        ylab("") +
        xlab("Subject") +
        theme_classic(base_size = 11, base_family = "sans") +
        theme(legend.text = element_markdown(), 
              legend.position = "right",
              axis.title.y = element_blank(),
              axis.text.x = element_blank()) +
        guides(fill=guide_legend(title="Gram-stain")) +
        scale_fill_manual(values = c(RColorBrewer::brewer.pal(n = 11, name = "Paired"))) +
        facet_wrap (~ treatment, scales = "free_x", nrow = 1) +
        labs(tag = "B")
        #facet_wrap (~ factor(sample_type, levels = c("Mock", "BAL", "Nasal", "Sputum")),
        #            scales= "free_x", nrow=2) +


b <- tax_table(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Neg.") %>%
        subset_samples(.,
                       baylor_other_id != "20220606_Neg") %>%
                transform_sample_counts(., function(x){ifelse(is.na(x), 0, x)})
        ) %>%
        cbind(species20 = "[Other]") %>%
        {top20species <- head(taxa_sums(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Neg.") %>%
                                      subset_samples(.,
                                                     baylor_other_id != "20220606_Neg") %>%
                                      transform_sample_counts(.,
                                                              function(x){ifelse(is.na(x),
                                                                                 0, 
                                                                                 x)})
                              ) %>%
                                data.frame %>%
                                arrange(-.) %>%
                                row.names(), 10)
   .[top20species, "species20"] <- as.character(.[top20species, "Species"])
   .[, 9] <- gsub("s__", " ", .[, 9])
   .[, 9] <- gsub("_", " ", .[, 9])
   .[, 9] <- gsub("[]]|[[]", "",  .[, 9])
   .[, 9] <- gsub(" sp", " sp.",  .[, 9])
   .[, 9] <- gsub(" sp.", "</i> sp.",  .[, 9])
   .[, 9] <- gsub(" group", "</i> group.",  .[, 9])
   .[, 9] <- ifelse(grepl("Other",.[, 9]),
                    "Other",
                    ifelse(grepl("</i>",  .[, 9]),
                           paste("<i>",  .[, 9], sep = ""),
                           paste("<i>",  .[, 9], "</i>", sep = "")) %>%
                            gsub("s__", "", .) %>%
                            gsub("_", " ", .)
   )
   phyloseq_temp <- subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Neg.") %>%
        subset_samples(.,
                       baylor_other_id != "20220606_Neg") %>%
                transform_sample_counts(., function(x){ifelse(is.na(x), 0, x)})
   tax_table(phyloseq_temp) <- tax_table(.) 
   phyloseq_temp
  } %>% 
        my_plot_bar(., fill="gram_stain") + 
        ylab("") +
        xlab("Subject") +
        theme_classic(base_size = 11, base_family = "sans") +
        theme(legend.text = element_markdown(),
              axis.text.x = element_blank()) +
        guides(fill=guide_legend(title="Gram-stain")) +
        scale_fill_manual(values = c(RColorBrewer::brewer.pal(n = 11, name = "Paired"))) +
        #facet_wrap (~ treatment, scales = "free_x", nrow = 1) +
        facet_wrap ( ~ treatment,
                    scales= "free_x", nrow=1) +
        ggtitle("E  Negative controls")

c <- tax_table(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "BAL") %>%
                transform_sample_counts(., function(x){ifelse(is.na(x), 0, x)})
        ) %>%
        cbind(species20 = "[Other]") %>%
        {top20species <- head(taxa_sums(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "BAL") %>%
                                      transform_sample_counts(.,
                                                              function(x){ifelse(is.na(x),
                                                                                 0, 
                                                                                 x)})
                              ) %>%
                                data.frame %>%
                                arrange(-.) %>%
                                row.names(), 10)
   .[top20species, "species20"] <- as.character(.[top20species, "Species"])
   .[, 9] <- gsub("s__", " ", .[, 9])
   .[, 9] <- gsub("_", " ", .[, 9])
   .[, 9] <- gsub("[]]|[[]", "",  .[, 9])
   .[, 9] <- gsub(" sp", " sp.",  .[, 9])
   .[, 9] <- gsub(" sp.", "</i> sp.",  .[, 9])
   .[, 9] <- gsub(" group", "</i> group.",  .[, 9])
   .[, 9] <- ifelse(grepl("Other",.[, 9]),
                    "Other",
                    ifelse(grepl("</i>",  .[, 9]),
                           paste("<i>",  .[, 9], sep = ""),
                           paste("<i>",  .[, 9], "</i>", sep = "")) %>%
                            gsub("s__", "", .) %>%
                            gsub("_", " ", .)
   )
   phyloseq_temp <- subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "BAL") %>%
           transform_sample_counts(., function(x){ifelse(is.na(x), 0, x)})
   tax_table(phyloseq_temp) <- tax_table(.) 
   sample_data(phyloseq_temp)$subject_id <- sample_data(phyloseq_temp)$subject_id %>% 
           factor(labels = c("A", "B", "C", "D", "E"))
   phyloseq_temp
  } %>% 
        my_plot_bar(., x = "treatment", fill="gram_stain") + 
        ylab("") +
        theme_classic(base_size = 11, base_family = "sans") +
        theme(legend.text = element_markdown(),
              axis.title.y = element_blank(),
              axis.text.x = element_text(angle = 45, hjust=1, size = 7,
                                         color = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3")),

              axis.title.x = element_blank()) +
        guides(fill=guide_legend(title="Gram-stain")) +
        scale_fill_manual(values = c(RColorBrewer::brewer.pal(n = 11, name = "Paired"))) +
        #facet_wrap (~ treatment, scales = "free_x", nrow = 1) +
        facet_wrap ( ~ subject_id,
                    scales= "free_x", nrow=1) +
        ggtitle("A  BAL")


d <- tax_table(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Nasal") %>% 
           subset_samples(., S.obs != 0)) %>%
        cbind(species20 = "[Other]") %>%
        {top20species <- head(taxa_sums(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Nasal") %>% 
           subset_samples(., S.obs != 0)) %>%
                                data.frame %>%
                                arrange(-.) %>%
                                row.names(), 10)
   .[top20species, "species20"] <- as.character(.[top20species, "Species"])
   .[, 9] <- gsub("s__", " ", .[, 9])
   .[, 9] <- gsub("_", " ", .[, 9])
   .[, 9] <- gsub("[]]|[[]", "",  .[, 9])
   .[, 9] <- gsub(" sp", " sp.",  .[, 9])
   .[, 9] <- gsub(" sp.", "</i> sp.",  .[, 9])
   .[, 9] <- gsub(" group", "</i> group.",  .[, 9])
   .[, 9] <- ifelse(grepl("Other",.[, 9]),
                    "Other",
                    ifelse(grepl("</i>",  .[, 9]),
                           paste("<i>",  .[, 9], sep = ""),
                           paste("<i>",  .[, 9], "</i>", sep = "")) %>%
                            gsub("s__", "", .) %>%
                            gsub("_", " ", .)
   )
   phyloseq_temp <- subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Nasal") %>% 
           subset_samples(., S.obs != 0)
   tax_table(phyloseq_temp) <- tax_table(.) 
   sample_data(phyloseq_temp)$subject_id <- sample_data(phyloseq_temp)$subject_id %>% 
           factor(labels = c("A", "B", "C", "D", "E", "F", "G", "H", "I", "J")) %>% 
           as.character()
   sample_data(phyloseq_temp)$treatment_order <- sample_data(phyloseq_temp)$treatment %>% 
           as.numeric() %>%
           as.character()
   phyloseq_temp
  } %>% 
        my_plot_bar(., x = "treatment_order", fill="gram_stain") + 
        theme_classic(base_size = 11, base_family = "sans") +
        theme(legend.position = "none",
              axis.title.x = element_blank(),
              axis.text.x = element_text(angle = 45, hjust=1, size = 7,
                                         color = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3")),
              axis.title.y = element_blank()) +
        guides(fill=guide_legend(title="Gram-stain")) +
        scale_fill_manual(values = c(RColorBrewer::brewer.pal(n = 11, name = "Paired"))) +
        scale_x_discrete(breaks = c("1", "2", "3", "4", "5", "6"),
                         labels = c("Untreated", "lyPMA", "Benzonase", "HostZero", "MolYsis", "QIAamp")) +
        ggtitle("B  Nasal swabs")

d <- ggplotGrob(d + facet_grid(~ subject_id, drop = T, scales = "free", space = "free"))



d$grobs[[22]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "lyPMA"     "QIAamp"

d$grobs[[23]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "lyPMA"     "QIAamp"

d$grobs[[22]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#fb9a99", "#a6cee3")
d$grobs[[23]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#fb9a99", "#a6cee3")

d$grobs[[24]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "Benzonase" "HostZero"  "MolYsis"

d$grobs[[25]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "Benzonase" "HostZero"  "MolYsis"

d$grobs[[26]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "Benzonase" "HostZero"  "MolYsis"

d$grobs[[27]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "Benzonase" "HostZero"  "MolYsis"

d$grobs[[24]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#33a02c", "#b2df8a", "#1f78b4")
d$grobs[[25]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#33a02c", "#b2df8a", "#1f78b4")
d$grobs[[26]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#33a02c", "#b2df8a", "#1f78b4")
d$grobs[[27]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#33a02c", "#b2df8a", "#1f78b4")



d$grobs[[28]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "lyPMA"     "Benzonase" "HostZero"

d$grobs[[28]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c","#fb9a99", "#33a02c", "#b2df8a")

d$grobs[[29]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "MolYsis"   "QIAamp"

d$grobs[[29]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#1f78b4", "#a6cee3")

d$grobs[[30]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "lyPMA"     "QIAamp"

d$grobs[[31]]$children[[2]]$grobs[[2]]$children[[1]]$label

## [1] "Untreated" "lyPMA"     "QIAamp"

d$grobs[[30]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#fb9a99", "#a6cee3")
d$grobs[[31]]$children[[2]]$grobs[[2]]$children[[1]]$gp$col <- c("#e31a1c", "#fb9a99", "#a6cee3")




e <- tax_table(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Sputum") %>% 
           subset_samples(., S.obs != 0)) %>%
        cbind(species20 = "[Other]") %>%
        {top20species <- head(taxa_sums(subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Sputum") %>% 
           subset_samples(., S.obs != 0)) %>%
                                data.frame %>%
                                arrange(-.) %>%
                                row.names(), 10)
   .[top20species, "species20"] <- as.character(.[top20species, "Species"])
   .[, 9] <- gsub("s__", " ", .[, 9])
   .[, 9] <- gsub("_", " ", .[, 9])
   .[, 9] <- gsub("[]]|[[]", "",  .[, 9])
   .[, 9] <- gsub(" sp", " sp.",  .[, 9])
   .[, 9] <- gsub(" sp.", "</i> sp.",  .[, 9])
   .[, 9] <- gsub(" group", "</i> group.",  .[, 9])
   .[, 9] <- ifelse(grepl("Other",.[, 9]),
                    "Other",
                    ifelse(grepl("</i>",  .[, 9]),
                           paste("<i>",  .[, 9], sep = ""),
                           paste("<i>",  .[, 9], "</i>", sep = "")) %>%
                            gsub("s__", "", .) %>%
                            gsub("_", " ", .)
   )
   phyloseq_temp <- subset_samples(phyloseq$phyloseq_rel,
                              sample_type == "Sputum") %>% 
           subset_samples(., S.obs != 0)
   tax_table(phyloseq_temp) <- tax_table(.) 
   sample_data(phyloseq_temp)$subject_id <- sample_data(phyloseq_temp)$subject_id %>% 
           factor(labels = c("A", "B", "C", "D", "E"))
   phyloseq_temp
  } %>%
        my_plot_bar(., x = "treatment", fill="gram_stain")+
        theme_classic(base_size = 11, base_family = "sans") +
        theme(legend.text = element_markdown(),
              axis.title.y = element_blank(),
              axis.text.x = element_text(angle = 45, hjust=1, size = 7, 
                                         color = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3")),

              axis.title.x = element_blank()) +
        guides(fill=guide_legend(title="Gram-stain")) +
        #scale_x_discrete(drop=F) +
        scale_fill_manual(values = c(RColorBrewer::brewer.pal(n = 11, name = "Paired"))) +
        facet_wrap ( ~ subject_id,
                    scales= "free_x", nrow=1) +
        #facet_wrap (~ factor(sample_type, levels = c("Mock", "BAL", "Nasal", "Sputum")),
        #            scales= "free_x", nrow=2) +
        ggtitle("C  Sputum")



figS6 <- ggarrange(c, d %>% as_ggplot, e, ncol = 1, common.legend = T, legend = "top")



annotate_figure(figS6,
                left = text_grob("Relative abundance",
                                 rot = 90,
                                 family = "sans", 
                                 size = 11),
                bottom = text_grob("Treatment",
                                 rot = 0,
                                 family = "sans", 
                                 size = 11)
)

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS6_updated.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 200, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue


annotate_figure(figS6,
                left = text_grob("Relative abundance",
                                 rot = 90,
                                 family = "sans", 
                                 size = 11),
                bottom = text_grob("Treatment",
                                 rot = 0,
                                 family = "sans", 
                                 size = 11)
)


# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Gram-stain stats (all samples)

Effect size, standard error (SE) and p-value at a statistical test on gram-negative proportion using linear mixed effect model. lmer( Gram-negative proportion vs sample_type + treatment + sample_type * treatment + (1|subject_id) )

Interaction term was significant (p-value = 0.018771)

Raw results

lmer_gram_stain_proportion <- lmer(gram_neg_prop ~ sample_type + treatment + sample_type * treatment + (1|subject_id),
     data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum")))

lmer_gram_stain_proportion %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: gram_neg_prop ~ sample_type + treatment + sample_type * treatment +  
##     (1 | subject_id)
##    Data: sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL",  
##     "Nasal", "Sputum"))
## 
## REML criterion at convergence: 699.8
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -3.2364 -0.2334 -0.0046  0.2513  2.4769 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 309.8    17.60   
##  Residual               231.6    15.22   
## Number of obs: 95, groups:  subject_id, 20
## 
## Fixed effects:
##                                      Estimate Std. Error      df t value
## (Intercept)                            32.848     10.406  36.974   3.157
## sample_typeNasal                      -32.005     12.744  36.974  -2.511
## sample_typeSputum                      28.794     14.716  36.974   1.957
## treatmentlyPMA                          6.416      9.624  61.541   0.667
## treatmentBenzonase                     15.332      9.624  61.541   1.593
## treatmentHostZERO                       8.609      9.624  61.541   0.895
## treatmentMolYsis                        3.372      9.624  61.541   0.350
## treatmentQIAamp                         7.306      9.624  61.541   0.759
## sample_typeNasal:treatmentlyPMA        13.747     13.087  63.733   1.050
## sample_typeSputum:treatmentlyPMA      -47.305     13.611  61.541  -3.476
## sample_typeNasal:treatmentBenzonase   -14.674     13.117  64.032  -1.119
## sample_typeSputum:treatmentBenzonase  -67.812     13.611  61.541  -4.982
## sample_typeNasal:treatmentHostZERO     -9.817     13.117  64.032  -0.748
## sample_typeSputum:treatmentHostZERO   -68.467     13.611  61.541  -5.030
## sample_typeNasal:treatmentMolYsis      -1.847     13.087  63.733  -0.141
## sample_typeSputum:treatmentMolYsis    -63.239     13.611  61.541  -4.646
## sample_typeNasal:treatmentQIAamp       -5.940     13.117  64.032  -0.453
## sample_typeSputum:treatmentQIAamp     -67.898     13.611  61.541  -4.989
##                                      Pr(>|t|)    
## (Intercept)                           0.00317 ** 
## sample_typeNasal                      0.01653 *  
## sample_typeSputum                     0.05797 .  
## treatmentlyPMA                        0.50747    
## treatmentBenzonase                    0.11627    
## treatmentHostZERO                     0.37451    
## treatmentMolYsis                      0.72725    
## treatmentQIAamp                       0.45066    
## sample_typeNasal:treatmentlyPMA       0.29749    
## sample_typeSputum:treatmentlyPMA      0.00094 ***
## sample_typeNasal:treatmentBenzonase   0.26746    
## sample_typeSputum:treatmentBenzonase 5.40e-06 ***
## sample_typeNasal:treatmentHostZERO    0.45697    
## sample_typeSputum:treatmentHostZERO  4.52e-06 ***
## sample_typeNasal:treatmentMolYsis     0.88821    
## sample_typeSputum:treatmentMolYsis   1.83e-05 ***
## sample_typeNasal:treatmentQIAamp      0.65221    
## sample_typeSputum:treatmentQIAamp    5.27e-06 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

ANOVA on lmer result

lmer_gram_stain_proportion %>% anova 

Table S5. Gram-stain stats -stratified

Stratified analsyis

Table S5. Effect size, standard error (SE) and p-value at a statistical test on gram-negative proportion using linear mixed effect model. LMER(Gram-negative proportion vs sample type + treatment + sample type * treatment + (1|subject id)). Stratified analyses were conducted for each sample type as an interaction term of sample type and treatment was significant at an ANOVA test (p-value < 0.001) using a model, LMER (Gram-negative proportion ~ sample type + treatment + sample type * treatment + (1|subject_id) ). The baseline of categorical variables is untreated BAL, and statistical significances were noted with *: p-value < 0.05 and ***: p-value < 0.001.

Raw results - Mock

##Mock
gram_neg_prop_mock <- 
        lm(gram_neg_prop ~ treatment,
     data = sample_data(phyloseq$phyloseq_rel) %>% data.frame %>% subset(., .$sample_type %in% c("Mock"))) 

gram_neg_prop_mock %>%
        summary

## 
## Call:
## lm(formula = gram_neg_prop ~ treatment, data = sample_data(phyloseq$phyloseq_rel) %>% 
##     data.frame %>% subset(., .$sample_type %in% c("Mock")))
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -11.2550  -0.0878  -0.0101   0.1161  17.2213 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          51.131      2.144  23.848  < 2e-16 ***
## treatmentlyPMA      -23.519      3.180  -7.396 9.54e-08 ***
## treatmentBenzonase  -51.082      3.180 -16.063 1.11e-14 ***
## treatmentHostZERO   -51.131      3.180 -16.078 1.08e-14 ***
## treatmentMolYsis    -50.638      3.180 -15.923 1.35e-14 ***
## treatmentQIAamp     -51.052      3.180 -16.054 1.12e-14 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5.252 on 25 degrees of freedom
## Multiple R-squared:  0.9497, Adjusted R-squared:  0.9397 
## F-statistic: 94.46 on 5 and 25 DF,  p-value: 2.085e-15

BAL

##BAL
gram_neg_prop_bal <- 
lmer(gram_neg_prop ~ treatment + (1|subject_id),
     data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL"))) 


gram_neg_prop_bal %>%
        summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: gram_neg_prop ~ treatment + (1 | subject_id)
##    Data: sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL"))
## 
## REML criterion at convergence: 240.8
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -2.07050 -0.39650  0.00736  0.38972  1.58848 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 1222.6   34.97   
##  Residual                575.6   23.99   
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error     df t value Pr(>|t|)
## (Intercept)          32.848     18.964  7.248   1.732    0.125
## treatmentlyPMA        6.416     15.174 20.000   0.423    0.677
## treatmentBenzonase   15.332     15.174 20.000   1.010    0.324
## treatmentHostZERO     8.609     15.174 20.000   0.567    0.577
## treatmentMolYsis      3.372     15.174 20.000   0.222    0.826
## treatmentQIAamp       7.306     15.174 20.000   0.481    0.635
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.400                            
## trtmntBnzns -0.400  0.500                     
## trtmntHZERO -0.400  0.500  0.500              
## trtmntMlYss -0.400  0.500  0.500  0.500       
## trtmntQIAmp -0.400  0.500  0.500  0.500  0.500

Nasal

##Nasal
gram_neg_prop_ns <- 
lmer(gram_neg_prop ~ treatment + (1|subject_id),
     data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal"))) 


gram_neg_prop_ns %>%
        summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: gram_neg_prop ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal"))
## 
## REML criterion at convergence: 191.3
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -3.1779 -0.2404 -0.0354  0.2250  2.9273 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept)  5.174   2.275   
##  Residual               25.784   5.078   
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                    Estimate Std. Error       df t value Pr(>|t|)    
## (Intercept)         0.84374    1.75949 26.89687   0.480    0.635    
## treatmentlyPMA     19.41142    2.84800 23.04516   6.816  5.9e-07 ***
## treatmentBenzonase  1.89039    2.85065 23.31583   0.663    0.514    
## treatmentHostZERO   0.02514    2.85065 23.31583   0.009    0.993    
## treatmentMolYsis    2.27684    2.84800 23.04516   0.799    0.432    
## treatmentQIAamp     0.13386    2.85065 23.31583   0.047    0.963    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.515                            
## trtmntBnzns -0.514  0.286                     
## trtmntHZERO -0.514  0.286  0.365              
## trtmntMlYss -0.515  0.272  0.349  0.349       
## trtmntQIAmp -0.514  0.349  0.269  0.269  0.286

Sputum

##Sputum
gram_neg_prop_spt <- 
lmer(gram_neg_prop ~ treatment + (1|subject_id),
     data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum"))) 

gram_neg_prop_spt %>%
        summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: gram_neg_prop ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum"))
## 
## REML criterion at convergence: 194
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -2.13570 -0.23729  0.08356  0.34246  2.24179 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept)  39.45    6.281  
##  Residual               103.99   10.198  
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          61.642      5.356  17.414  11.509 1.44e-09 ***
## treatmentlyPMA      -40.888      6.450  20.000  -6.340 3.46e-06 ***
## treatmentBenzonase  -52.480      6.450  20.000  -8.137 8.96e-08 ***
## treatmentHostZERO   -59.857      6.450  20.000  -9.281 1.09e-08 ***
## treatmentMolYsis    -59.867      6.450  20.000  -9.282 1.09e-08 ***
## treatmentQIAamp     -60.592      6.450  20.000  -9.395 8.94e-09 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.602                            
## trtmntBnzns -0.602  0.500                     
## trtmntHZERO -0.602  0.500  0.500              
## trtmntMlYss -0.602  0.500  0.500  0.500       
## trtmntQIAmp -0.602  0.500  0.500  0.500  0.500

Tidy summarized table

gram_neg_prop_mock_kbl <- gram_neg_prop_mock %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              gram_neg_prop_mock %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]

gram_neg_prop_ns_kbl  <- gram_neg_prop_ns %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              gram_neg_prop_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


gram_neg_prop_spt_kbl <- gram_neg_prop_spt %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              gram_neg_prop_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]



gram_neg_prop_bal_kbl <- gram_neg_prop_bal %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              gram_neg_prop_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


tableS5 <- cbind(gram_neg_prop_mock_kbl, gram_neg_prop_bal_kbl, gram_neg_prop_ns_kbl, gram_neg_prop_spt_kbl) %>%
        kbl(format = "html", escape = 0) %>% kable_styling(full_width = 0, html_font = "sans") %>% 
        add_header_above(c(" " = 1, "Mock" = 3, "BAL" = 3, "Nasal swab" = 3, "Sputum" = 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans") 

tableS5

save_kable(tableS5, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS5.html", self_contained = T)

Fig. 3 Alpha and beta diversity

Fig. 3. Alpha and beta diversity by sample type and treatment method after removing potential contaminants and rare taxa. (A) Species richness with statistical test results (linear mixed effect model stratified by sample type), (B) Morisita-Horn dissimilarity within subject between treatment, representing squares for median value and bars for 95% confidence intervals.

• PSL comments (20230630): - Figure 3A: it is hard to see some of the boxplot colors due to the narrow interquartile range. Can you use stat_summary and geom = “pointrange” like what Maghini DG et al did for their Figure 3b or create dotplot + pointrange geom like their Figure 3c? I found some example code goodgling though you might want to show median + iqr rather than mean + sd stat_summary(fun = mean, geom = “pointrange”, fun.max = function(x) mean(x) + sd(x), fun.min = function(x) mean(x) - sd(x))
  • Figure 3A: we also need to talk because the mock community is only supposed to have a species richness = 10. I know that some of this may be due to taxonomic misclassification but that immediately will raise issues with the reviewer so we need to anticipate how to address this in advance
  • Figure 3A: add relevant p-values using horizontal bars and stars for significance like Maghini DG et al did for Figure 3B
  • Figure 3A: why not just do what Maghini DG et al did for Figure 3e? But have one panel per sample type

f3a <- ggplot(subset(sample_data(phyloseq$phyloseq_count) %>% 
                             data.frame, sample_data(phyloseq$phyloseq_count)$sample_type %in% c(#"Mock",
                                     "Sputum", "Nasal", "BAL")), aes(x = treatment, y = S.obs)) +
        geom_jitter(aes(color = treatment),
                    position = position_jitter(0.2), size = 1.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        #stat_summary(fun = mean,
        #       geom = "pointrange",
        #       fun.max = function(x) mean(x) + sd(x),
        #       fun.min = function(x) mean(x) - sd(x))+
        ylab("Species richness") +
        xlab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using 
        scale_fill_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using 
        labs(tag = "A") +
        theme(plot.tag = element_text(size = 15),
              axis.text.x = element_blank(),
              axis.ticks.x = element_blank(),
              legend.position = "top") +
        facet_wrap(~sample_type, nrow = 1) + 
        guides(col = guide_legend(nrow = 1)) 

dat_text <- data.frame(
  label = c(
          #"", "***", "***", "**", "***", #label for Mock
          "", "", "", "*", "", #label for BAL
          "", "", "***", "*", "**", 
          "**", "***", "***", "***", "***"),
  sample_type = c(
          #"Mock", "Mock", "Mock", "Mock", "Mock", 
          "BAL", "BAL", "BAL", "BAL", "BAL", 
          "Nasal", "Nasal", "Nasal", "Nasal", "Nasal", 
          "Sputum", "Sputum", "Sputum", "Sputum", "Sputum"),
  treatment     = c(
          #"lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp", 
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
  S.obs = c(
          #50, 30, 35, 33, 31,
          30, 35, 50, 52, 50,
          30, 30, 30, 35, 30,
          100, 120, 147, 140, 125)
)



dat_text$sample_type <- factor(dat_text$sample_type, levels = c("BAL", "Nasal", "Sputum"
                                                                ))
dat_text$treatment <- factor(dat_text$treatment, levels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))


f3a <- f3a + geom_text(
  data    = dat_text,
  mapping = aes(x = treatment, y = S.obs, label = label)
)



f5S_mock <- ggplot(subset(sample_data(phyloseq$phyloseq_count) %>% 
                             data.frame,
                          sample_data(phyloseq$phyloseq_count)$sample_type %in%
                                  c("Mock")),
                   aes(x = treatment, y = S.obs)) +
        geom_jitter(aes(color = treatment), position = position_jitter(0.2),
                    size = 1.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        ylab("Species richness") +
        xlab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using 
        labs(tag = "C") +
        theme(plot.tag = element_text(size = 15),
              axis.text.x = element_blank(),
              axis.ticks.x = element_blank(),
              legend.position = "top") +
        facet_wrap(~sample_type, nrow = 1) + 
        guides(col = guide_legend(nrow = 1)) 

dat_text <- data.frame(
  label = c(
          "", "***", "***", "**", "***"#, #label for Mock
          #"", "", "", "*", "", #label for BAL
          #"", "", "***", "*", "**", 
          #"**", "***", "***", "***", "***"
          ),
  sample_type = c(
          "Mock", "Mock", "Mock", "Mock", "Mock"#, 
          #"BAL", "BAL", "BAL", "BAL", "BAL", 
          #"Nasal", "Nasal", "Nasal", "Nasal", "Nasal", 
          #"Sputum", "Sputum", "Sputum", "Sputum", "Sputum"
          ),
  treatment = c(
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"#, 
          #"lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          #"lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          #"lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"
          ),
  S.obs = c(
          50, 30, 35, 33, 31
          #30, 35, 50, 52, 50,
          #30, 30, 30, 35, 30,
          #100, 120, 147, 140, 125
          )
)



dat_text$sample_type <- factor(dat_text$sample_type, levels = c("Mock"#, 
                                                                #"BAL", "Nasal", "Sputum"
                                                                ))
dat_text$treatment <- factor(dat_text$treatment, levels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))


f5S_mock <- f5S_mock + geom_text(
  data    = dat_text,
  mapping = aes(x = treatment, y = S.obs, label = label)
)



#Making subset of non-zero samples without neg

phyloseq_rel_nz <- phyloseq$phyloseq_rel %>%
        subset_samples(S.obs != 0 & sample_type %in% c("Mock", 
                                                       "BAL", "Nasal", "Sputum"))

#distances of betadiversity - boxplots
horn_dist_long <- distance(phyloseq_rel_nz, method="horn") %>% as.matrix() %>% melt_dist() #making long data of distance matrices

#Adding sample type and treatment name. 
#this can be also done by merging metadata into the `horn_dist_long`
names <- data.frame(str_split_fixed(horn_dist_long$iso1, "_", 3))
names2 <- data.frame(str_split_fixed(horn_dist_long$iso2, "_", 3))
horn_dist_long$sample_id_1 <- paste(names$X1, names$X2, sep = "_")
horn_dist_long$method_1 <- ifelse(grepl("lyPMA", horn_dist_long$iso1),"lypma", 
                                         ifelse(grepl("ben", horn_dist_long$iso1),"benzonase", 
                                                ifelse(grepl("host", horn_dist_long$iso1),"host_zero", 
                                                       ifelse(grepl("qia", horn_dist_long$iso1),"qiaamp", 
                                                              ifelse(grepl("moly", horn_dist_long$iso1),"molysis", 
                                                                     "control")))))


#Adding data for iso 2 also should be done
horn_dist_long$sample_id_2 <- paste(names2$X1, names2$X2, sep = "_")
horn_dist_long$method_2 <-ifelse(grepl("lyPMA", horn_dist_long$iso2),"lypma", 
                                        ifelse(grepl("ben", horn_dist_long$iso2),"benzonase", 
                                               ifelse(grepl("host", horn_dist_long$iso2),"host_zero", 
                                                      ifelse(grepl("qia", horn_dist_long$iso2),"qiaamp", 
                                                             ifelse(grepl("moly", horn_dist_long$iso2),"molysis", 
                                                                    "control")))))


#subsetting distances of my interest
horn_dist_long$sample_id_1 <- ifelse(grepl("pos", horn_dist_long$sample_id_1, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", horn_dist_long$sample_id_1, ignore.case = T),"Neg.",
                                        horn_dist_long$sample_id_1))
horn_dist_long$sample_id_2 <- ifelse(grepl("pos", horn_dist_long$sample_id_2, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", horn_dist_long$sample_id_2, ignore.case = T),"Neg.",
                                        horn_dist_long$sample_id_2))


horn_dist_long_within_sampleid_from_control <- subset(horn_dist_long, horn_dist_long$sample_id_1 == horn_dist_long$sample_id_2) # data within samples

horn_dist_long_within_sampleid_from_control <- subset(horn_dist_long_within_sampleid_from_control,
                                                           horn_dist_long_within_sampleid_from_control$method_1 != horn_dist_long_within_sampleid_from_control$method_2) # remove irrelevant association

horn_dist_long_within_sampleid_from_control <- subset(horn_dist_long_within_sampleid_from_control, (horn_dist_long_within_sampleid_from_control$method_1 == "control") + (horn_dist_long_within_sampleid_from_control$method_2 == "control") != 0)


horn_dist_long_within_sampleid_from_control$treatment <- horn_dist_long_within_sampleid_from_control$method_1

horn_dist_long_within_sampleid_from_control$treatment <- ifelse(horn_dist_long_within_sampleid_from_control$treatment == "control", horn_dist_long_within_sampleid_from_control$method_2, horn_dist_long_within_sampleid_from_control$treatment) 


#Setting key method
horn_dist_long_within_sampleid_from_control$sample_type <- ifelse(grepl("NS", horn_dist_long_within_sampleid_from_control$iso1), "Nasal",
                                                                  ifelse(grepl("CFB", horn_dist_long_within_sampleid_from_control$iso1), "Sputum",
                                                                         ifelse(grepl("BAL", horn_dist_long_within_sampleid_from_control$iso1), "BAL",
                                                                                ifelse(grepl("pos|POS", horn_dist_long_within_sampleid_from_control$iso1, ignore.case = T), "Mock",
                                                                                       ifelse(grepl("neg|N_EXT", horn_dist_long_within_sampleid_from_control$iso1), "Neg.",NA)))))

#Making a column for baseline (controls, from where?)
horn_dist_long_within_sampleid_from_control <- horn_dist_long_within_sampleid_from_control %>% 
        mutate(dist_from = case_when(method_1 == "control" ~ iso1,
                                     method_2 == "control" ~ iso2))

dummy <- data.frame(iso1 = horn_dist_long_within_sampleid_from_control$dist_from %>% unique,
           iso2 = horn_dist_long_within_sampleid_from_control$dist_from %>% unique,
           dist = 0,
           treatment = "Untreated",
           method_1 = "control",
           method_2 = "control"
           )
names <- data.frame(str_split_fixed(dummy$iso1, "_", 3))
names2 <- data.frame(str_split_fixed(dummy$iso2, "_", 3))
dummy$sample_id_1 <- paste(names$X1, names$X2, sep = "_")
#Adding data for iso 2 also should be done
dummy$sample_id_2 <- paste(names2$X1, names2$X2, sep = "_")


#subsetting distances of my interest
dummy$sample_id_1 <- ifelse(grepl("pos", dummy$sample_id_1, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", dummy$sample_id_1, ignore.case = T),"Neg.",
                                        dummy$sample_id_1))
dummy$sample_id_2 <- ifelse(grepl("pos", dummy$sample_id_2, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", dummy$sample_id_2, ignore.case = T),"Neg.",
                                        dummy$sample_id_2))
dummy$sample_type <- ifelse(grepl("NS", dummy$iso1), "Nasal",
                            ifelse(grepl("CFB", dummy$iso1), "Sputum",
                                   ifelse(grepl("BAL", dummy$iso1), "BAL",
                                          ifelse(grepl("pos|POS", dummy$iso1, ignore.case = T), "Mock",
                                                 ifelse(grepl("neg|N_EXT", dummy$iso1), "Neg.",NA)))))
dummy <- subset(dummy, !is.na(dummy$sample_type))
horn_dist_long_within_sampleid_from_control <- bind_rows(horn_dist_long_within_sampleid_from_control, dummy)

#Here, sample id is the same as subject id.
horn_dist_long_within_sampleid_from_control$subject_id <- horn_dist_long_within_sampleid_from_control$sample_id_1

horn_dist_long_within_sampleid_from_control$treatment <-
        factor(horn_dist_long_within_sampleid_from_control$treatment,
               levels = c("Untreated", "lypma", "benzonase", "host_zero", "molysis", "qiaamp"))

#Making figure of beta diversity distances

## This only includes samples (BAL, Nasal and Sputum)
f3b2 <- rbind(cbind("Sputum",
            lmer(dist ~ treatment + (1|subject_id),
                data = horn_dist_long_within_sampleid_from_control %>%
                        subset(sample_type == "Sputum")) %>% 
                    confint()
            ),
      cbind("BAL",
            lmer(dist ~ treatment + (1|subject_id),
                 data = horn_dist_long_within_sampleid_from_control %>%
                         subset(sample_type == "BAL")) %>% 
                    confint()
            ),
      cbind("Nasal",
            lmer(dist ~ treatment + (1|subject_id),
                 data = horn_dist_long_within_sampleid_from_control %>%
                         subset(sample_type == "Nasal")) %>% 
                    confint()
     )
) %>% {
        row_names <- rownames(.)
        data_frame <- data.frame(.)
        data_frame$treatment <- row_names
        data_frame %>% 
                remove_rownames() %>%
                rename(sample_type = "V1",
               "2.5%" = "X2.5..",
               "97.5%" = "X97.5..") %>% 
                mutate(`2.5%` = as.numeric(`2.5%`),
                       `97.5%` = as.numeric(`97.5%`),
                       mean = (`97.5%`+`2.5%`)/2,
                       treatment = case_when(treatment == "(Intercept)" ~ "Untreated",
                                             treatment == "treatmentlypma" ~ "lyPMA",
                                             treatment == "treatmentbenzonase" ~ "Benzonase",
                                             treatment == "treatmenthost_zero" ~ "HostZERO",
                                             treatment == "treatmentmolysis" ~ "MolYsis",
                                             treatment == "treatmentqiaamp" ~ "QIAamp"
                                             ),
                       treatment = factor(treatment,
                                          levels = c("Untreated", "lyPMA", "Benzonase",
                                      "HostZERO", "MolYsis", "QIAamp")),
                       sample_type = factor(sample_type,
                                          levels = c("BAL", "Nasal", "Sputum"))
                       ) %>%
                subset(treatment %in% c(#"Untreated", 
                                        "lyPMA", "Benzonase",
                                      "HostZERO", "MolYsis", "QIAamp"))
} %>%
        ggplot(aes(x = mean, y = treatment, col = treatment)) +
        geom_point(aes(x=mean), shape=15, size=3) +
        geom_linerange(aes(xmin=`2.5%`, xmax=`97.5%`)) +
        facet_wrap(~sample_type, nrow = 4) +
        scale_y_discrete(limits=rev) +
        scale_color_manual(values = c(#"#e31a1c",
                                      "#fb9a99","#33a02c",
                                      "#b2df8a","#1f78b4","#a6cee3"),
                           name = "Treatment",
                           breaks = c(#"Untreated", 
                                      "lyPMA", "Benzonase",
                                      "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        xlab("Morisita-Horn dissimilarity from untreated") +
        ylab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        theme(plot.tag = element_text(size = 15),
              axis.text.y = element_blank(),
              axis.ticks.y = element_blank(),
              legend.position = "none") +
        labs(tag = "B") +
        geom_vline(xintercept = 0, col = "black", linetype="dotted") +
        #coord_cartesian(xlim=c(-0.5, 1)) +
        #geom_text(aes(x = 0, label = treatment), hjust = 0, nudge_x = -.55, size = 3, color = "black", family = "sans") +
        #geom_text(aes(x = 0, label = text), hjust = 0, nudge_x = -0.4, size = 3, color = "black", family = "sans") +
        scale_x_continuous(breaks = c(-0.25, 0, 0.25, 0.5, 0.75),
                           labels = c(-0.25, "0 (low bias)", 0.25, 0.5, "0.75 (high bias)"))


dat_text <- data.frame(
  label = c(
          #"", "***", "***", "**", "***", #label for Mock
          "*", "", "", "", "", #label for BAL
          "*", "*", "", "**", "", 
          "**", "***", "***", "***", "***"),
  sample_type = c(
          #"Mock", "Mock", "Mock", "Mock", "Mock", 
          "BAL", "BAL", "BAL", "BAL", "BAL", 
          "Nasal", "Nasal", "Nasal", "Nasal", "Nasal", 
          "Sputum", "Sputum", "Sputum", "Sputum", "Sputum"),
  treatment     = c(
          #"lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp", 
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
  mean = c(
          #50, 30, 35, 33, 31,
          0.6, 0, 0, 0, 0,
          0.32, 0.3, 0, 0.37, 0,
          0.58, 0.75, 0.85, 0.83, 0.85)
)



dat_text$sample_type <- factor(dat_text$sample_type, levels = c("BAL", "Nasal", "Sputum"
                                                                ))
dat_text$treatment <- factor(dat_text$treatment, levels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))


f3b2 <- f3b2 + geom_text(
  data    = dat_text,
  mapping = aes(y = treatment, x = mean, label = label),
  col = "black"
)


f3b2_box <- horn_dist_long_within_sampleid_from_control %>% 
        mutate(across(sample_type, factor, levels=c(#"Mock", 
                                                    "BAL", "Nasal","Sputum"
                                                    ))) %>%
        subset(., .$sample_type != "Neg.") %>% 
        subset(., .$treatment != "Untreated") %>%
  mutate(treatment = factor(treatment, levels = c("lypma", "benzonase", "host_zero", "molysis", "qiaamp"),
                            labels = c("lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))) %>%
        ggplot(aes(x = dist, y = treatment, col = treatment)) +
        geom_boxplot(aes(x=dist)) +
        #geom_linerange(aes(xmin=`2.5%`, xmax=`97.5%`)) +
        facet_wrap(~sample_type, nrow = 4) +
        scale_y_discrete(limits=rev) +
        scale_color_manual(values = c(#"#e31a1c",
                                      "#fb9a99","#33a02c",
                                      "#b2df8a","#1f78b4","#a6cee3"),
                           name = "Treatment",
                           breaks = c(#"Untreated", 
                                      "lyPMA", "Benzonase",
                                      "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        xlab("Morisita-Horn dissimilarity from untreated") +
        ylab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        theme(plot.tag = element_text(size = 15),
              axis.text.y = element_blank(),
              axis.ticks.y = element_blank(),
              legend.position = "none") +
        labs(tag = "B") +
        geom_vline(xintercept = 0, col = "black", linetype="dotted") +
        #coord_cartesian(xlim=c(-0.5, 1)) +
        #geom_text(aes(x = 0, label = treatment), hjust = 0, nudge_x = -.55, size = 3, color = "black", family = "sans") +
        #geom_text(aes(x = 0, label = text), hjust = 0, nudge_x = -0.4, size = 3, color = "black", family = "sans") +
        scale_x_continuous(breaks = c(-0.25, 0, 0.25, 0.5, 0.75),
                           labels = c(-0.25, "0 (low bias)", 0.25, 0.5, "0.75 (high bias)"))


dat_text <- data.frame(
  label = c(
          #"", "***", "***", "**", "***", #label for Mock
          "*", "", "", "", "", #label for BAL
          "*", "*", "", "**", "", 
          "**", "***", "***", "***", "***"),
  sample_type = c(
          #"Mock", "Mock", "Mock", "Mock", "Mock", 
          "BAL", "BAL", "BAL", "BAL", "BAL", 
          "Nasal", "Nasal", "Nasal", "Nasal", "Nasal", 
          "Sputum", "Sputum", "Sputum", "Sputum", "Sputum"),
  treatment     = c(
          #"lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp", 
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
  dist = c(
          #50, 30, 35, 33, 31,
          0.9, 0, 0, 0, 0,
          0.48, 0.4, 0, 0.55, 0,
          0.8, 0.82,1, 1.02, 0.98)
)



dat_text$sample_type <- factor(dat_text$sample_type, levels = c("BAL", "Nasal", "Sputum"
                                                                ))
dat_text$treatment <- factor(dat_text$treatment, levels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))


f3b2_box <- f3b2_box + geom_text(
  data    = dat_text,
  mapping = aes(y = treatment, x = dist, label = label),
  col = "black"
)



# Mock community's beta-diversity plot.
## This only includes positive control samples (Mock)

f5S_mock_mh <- horn_dist_long_within_sampleid_from_control %>% 
        mutate(across(sample_type, factor, levels=c("Mock"#, 
                                                    #"BAL", "Nasal","Sputum"
                                                    ))) %>%
        subset(., .$sample_type != "Neg.") %>% 
        group_by(sample_type, treatment) %>%
        summarise(mean = mean(dist, na.rm = TRUE),
            sd = sd(dist, na.rm = TRUE),
            n = n()) %>%
        subset(., .$treatment != "Untreated") %>%
  mutate(se = sd / sqrt(n),
         lower.ci = mean - qt(1 - (0.05 / 2), n - 1) * se,
         upper.ci = mean + qt(1 - (0.05 / 2), n - 1) * se,
         treatment = factor(treatment, levels = c(#"Untreated",
                                                  "lypma", "benzonase", "host_zero", "molysis", "qiaamp"),
                            labels = c(#"Untreated",
                                       "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))) %>%
        #,
         #text = paste(sprintf("%.2f", round(mean, digits = 2)), " [", sprintf("%.2f", round(lower.ci, digits = 2)), ", ", sprintf("%.2f", round(upper.ci, digits = 2)), "]", sep = "")) %>%
        ggplot(aes(x = mean, y = treatment, col = treatment)) +
        geom_point(aes(x=mean), shape=15, size=3) +
        geom_linerange(aes(xmin=lower.ci, xmax=upper.ci)) +
        facet_wrap(~sample_type, nrow = 4) +
        scale_y_discrete(limits=rev) +
        scale_color_manual(values = c(#"#e31a1c", 
                                      "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", 
                           labels = c(#"Untreated",
                                      "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        xlab("Morisita-Horn dissimilarity from untreated") +
        ylab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        theme(plot.tag = element_text(size = 15),
              axis.text.y = element_blank(),
              axis.ticks.y = element_blank(),
              legend.position = "none") +
        labs(tag = "D") +
        geom_vline(xintercept = 0, col = "black", linetype="dotted") +
        #coord_cartesian(xlim=c(-0.5, 1)) +
        #geom_text(aes(x = 0, label = treatment), hjust = 0, nudge_x = -.55, size = 3, color = "black", family = "sans") +
        #geom_text(aes(x = 0, label = text), hjust = 0, nudge_x = -0.4, size = 3, color = "black", family = "sans") +
        scale_x_continuous(breaks = c(-0.5, 0, 0.5, 1, 1.5), labels = c(-0.5, "0 (low bias)", 0.5, 1, "1.5 (high bias)"))



f5S_mock_mh_box <- horn_dist_long_within_sampleid_from_control %>% 
        mutate(across(sample_type, factor, levels=c("Mock"#, 
                                                    #"BAL", "Nasal","Sputum"
                                                    ))) %>%
        subset(., .$sample_type != "Neg.") %>% 
        subset(., .$treatment != "Untreated") %>%
  mutate(treatment = factor(treatment, levels = c(#"Untreated",
                                                  "lypma", "benzonase", "host_zero", "molysis", "qiaamp"),
                            labels = c(#"Untreated",
                                       "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))) %>%
        #,
         #text = paste(sprintf("%.2f", round(mean, digits = 2)), " [", sprintf("%.2f", round(lower.ci, digits = 2)), ", ", sprintf("%.2f", round(upper.ci, digits = 2)), "]", sep = "")) %>%
        ggplot(aes(x = dist, y = treatment, col = treatment)) +
        geom_boxplot() +
        facet_wrap(~sample_type, nrow = 4) +
        scale_y_discrete(limits=rev) +
        scale_color_manual(values = c(#"#e31a1c", 
                                      "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", 
                           labels = c(#"Untreated",
                                      "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        xlab("Morisita-Horn dissimilarity from untreated") +
        ylab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        theme(plot.tag = element_text(size = 15),
              axis.text.y = element_blank(),
              axis.ticks.y = element_blank(),
              legend.position = "none") +
        labs(tag = "D") +
        geom_vline(xintercept = 0, col = "black", linetype="dotted") +
        #coord_cartesian(xlim=c(-0.5, 1)) +
        #geom_text(aes(x = 0, label = treatment), hjust = 0, nudge_x = -.55, size = 3, color = "black", family = "sans") +
        #geom_text(aes(x = 0, label = text), hjust = 0, nudge_x = -0.4, size = 3, color = "black", family = "sans") +
        scale_x_continuous(breaks = c(-0.5, 0, 0.5, 1, 1.5), labels = c(-0.5, "0 (low bias)", 0.5, 1, "1.5 (high bias)"))


fig3 <- ggarrange(f3a, f3b2_box, ncol = 1, common.legend = T, align = "hv")

fig3

# png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/Figure3.png",   # The directory you want to save the file in
#     width = 180, # The width of the plot in inches
#     height = 220, # The height of the plot in inches
#     units = "mm",
#     res = 600
# ) #fixing multiple page issue
# 
# fig3
# # alpha diversity plots
# #ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
# #          ncol = 1) # alpha diversity plots
# 
# dev.off()



pdf(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/Figure3.pdf",   # The directory you want to save the file in
    width = 183 / 25.4, # Convert width from mm to inches (180 mm)
    height = 220 / 25.4, # Convert height from mm to inches (160 mm)
    paper = "special",   # Prevents default paper size settings
    onefile = FALSE      # Ensures a single page per PDF file
)


fig3
# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Calculation for log centered Final reads

Distribution of centralized final reads

sample_data <- sample_data(phyloseq$phyloseq_count)
sample_data$log_centered_final_reads <- log(sample_data$Final_reads + 1) - median(log((subset(sample_data, sample_data$sample_type %in% c("BAL") & sample_data$treatment %in% c("Untreated")) %>% .$Final_reads) + 1))

sample_data$bal_log_centered_final_reads <- log(sample_data$Final_reads + 1) - median(log((subset(sample_data, sample_data$sample_type %in% c("BAL") & sample_data$treatment %in% c("Untreated"))%>% .$Final_reads) + 1))

sample_data$ns_log_centered_final_reads <- log(sample_data$Final_reads + 1) - median(log((subset(sample_data, sample_data$sample_type %in% c("Nasal") & sample_data$treatment %in% c("Untreated"))%>% .$Final_reads) + 1))

sample_data$spt_log_centered_final_reads <- log(sample_data$Final_reads + 1) - median(log((subset(sample_data, sample_data$sample_type %in% c("Sputum") & sample_data$treatment %in% c("Untreated"))%>% .$Final_reads) + 1))


subset(sample_data, sample_data$sample_type %in% c("BAL", "Nasal", "Sputum")) %>% .$log_centered_final_reads %>% hist (main = "Histogram of centered log10 final reads of BAL, Nasal, Sputum")

subset(sample_data, sample_data$sample_type %in% c("BAL")) %>% .$log_centered_final_reads %>% hist (main = "Histogram of centered log10 final reads of BAL")

subset(sample_data, sample_data$sample_type %in% c("Nasal")) %>% .$log_centered_final_reads %>% hist (main = "Histogram of centered log10 final reads of Nasal")

subset(sample_data, sample_data$sample_type %in% c("Sputum")) %>% .$log_centered_final_reads %>% hist (main = "Histogram of centered log10 final reads of Sputum")

Species richness (all sample)

This table is too redundant. Removed from the manuscript..

Effect size, standard error (SE) and p-value of a statistical test for species richness with an interaction term using linear mixed effect model (Species richness ~ sample_type * treatment + log10 (Final_reads) + (1|subject_id) ).

Interaction term was highly significant (p = 2.2e-16)

Raw result

lmer_sr <- lmer(S.obs ~ treatment * sample_type + log10(Final_reads) + (1|subject_id),
                    data = sample_data %>% 
                            data.frame %>% 
                            subset(., .$S.obs != 0))

lmer_sr %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: 
## S.obs ~ treatment * sample_type + log10(Final_reads) + (1 | subject_id)
##    Data: sample_data %>% data.frame %>% subset(., .$S.obs != 0)
## 
## REML criterion at convergence: 933.1
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -4.0347 -0.2709  0.0045  0.2423  3.0128 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 96.87    9.842   
##  Residual               53.26    7.298   
## Number of obs: 155, groups:  subject_id, 22
## 
## Fixed effects:
##                                      Estimate Std. Error       df t value
## (Intercept)                          -72.0854    14.4625  53.4034  -4.984
## treatmentlyPMA                        -2.2692     4.4631 108.5096  -0.508
## treatmentBenzonase                    -2.7922     4.4207 108.3433  -0.632
## treatmentHostZERO                     -4.1979     4.4402 108.4205  -0.945
## treatmentMolYsis                      -2.7594     4.4258 108.3638  -0.623
## treatmentQIAamp                       -3.1897     4.4276 108.3708  -0.720
## sample_typeMock                        8.0057    14.9720  18.7520   0.535
## sample_typeBAL                         5.5843    11.8180  20.2523   0.473
## sample_typeNasal                      -3.2587    11.0408  18.4140  -0.295
## sample_typeSputum                     11.2252    11.6526  19.2014   0.963
## log10(Final_reads)                    13.1570     1.7246 116.2668   7.629
## treatmentlyPMA:sample_typeMock         7.0987     6.5463 109.1116   1.084
## treatmentBenzonase:sample_typeMock   -13.4773     6.2512 108.3417  -2.156
## treatmentHostZERO:sample_typeMock    -11.8560     6.2612 108.3698  -1.894
## treatmentMolYsis:sample_typeMock      -9.2558     6.2557 108.3544  -1.480
## treatmentQIAamp:sample_typeMock      -13.6156     6.2504 108.3393  -2.178
## treatmentlyPMA:sample_typeBAL         -1.2644     6.7944 108.3380  -0.186
## treatmentBenzonase:sample_typeBAL     -3.0508     6.7836 109.2406  -0.450
## treatmentHostZERO:sample_typeBAL      -0.2777     6.7685 109.1960  -0.041
## treatmentMolYsis:sample_typeBAL        7.1574     6.8419 109.4064   1.046
## treatmentQIAamp:sample_typeBAL        -1.9740     6.8397 109.4004  -0.289
## treatmentlyPMA:sample_typeNasal        4.6417     6.3470 112.1444   0.731
## treatmentBenzonase:sample_typeNasal    0.4571     6.1682 112.2096   0.074
## treatmentHostZERO:sample_typeNasal     2.8911     6.2602 112.5293   0.462
## treatmentMolYsis:sample_typeNasal      6.1367     6.1502 111.7217   0.998
## treatmentQIAamp:sample_typeNasal      -2.7832     6.3453 112.1047  -0.439
## treatmentlyPMA:sample_typeSputum      32.7634     6.3974 108.3573   5.121
## treatmentBenzonase:sample_typeSputum  58.2631     6.5294 108.7077   8.923
## treatmentHostZERO:sample_typeSputum   85.1820     6.8449 109.4565  12.445
## treatmentMolYsis:sample_typeSputum    89.3757     7.1413 110.0631  12.515
## treatmentQIAamp:sample_typeSputum     69.7623     6.7477 109.2382  10.339
##                                      Pr(>|t|)    
## (Intercept)                          6.92e-06 ***
## treatmentlyPMA                         0.6122    
## treatmentBenzonase                     0.5290    
## treatmentHostZERO                      0.3465    
## treatmentMolYsis                       0.5343    
## treatmentQIAamp                        0.4728    
## sample_typeMock                        0.5991    
## sample_typeBAL                         0.6416    
## sample_typeNasal                       0.7712    
## sample_typeSputum                      0.3474    
## log10(Final_reads)                   7.12e-12 ***
## treatmentlyPMA:sample_typeMock         0.2806    
## treatmentBenzonase:sample_typeMock     0.0333 *  
## treatmentHostZERO:sample_typeMock      0.0609 .  
## treatmentMolYsis:sample_typeMock       0.1419    
## treatmentQIAamp:sample_typeMock        0.0315 *  
## treatmentlyPMA:sample_typeBAL          0.8527    
## treatmentBenzonase:sample_typeBAL      0.6538    
## treatmentHostZERO:sample_typeBAL       0.9674    
## treatmentMolYsis:sample_typeBAL        0.2978    
## treatmentQIAamp:sample_typeBAL         0.7734    
## treatmentlyPMA:sample_typeNasal        0.4661    
## treatmentBenzonase:sample_typeNasal    0.9411    
## treatmentHostZERO:sample_typeNasal     0.6451    
## treatmentMolYsis:sample_typeNasal      0.3205    
## treatmentQIAamp:sample_typeNasal       0.6618    
## treatmentlyPMA:sample_typeSputum     1.33e-06 ***
## treatmentBenzonase:sample_typeSputum 1.24e-14 ***
## treatmentHostZERO:sample_typeSputum   < 2e-16 ***
## treatmentMolYsis:sample_typeSputum    < 2e-16 ***
## treatmentQIAamp:sample_typeSputum     < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

ANOVA result on LMER result for species richness

lmer_sr %>% anova()

Table S3. Species richness (stratified) without depth

Table S3. Effect size, standard error (SE) and p-value of a statistical test for species richness using linear mixed effect model stratified by sample type (Species richness ~ treatment + (1|subject_id) ). Stratified analyses were conducted for each sample type as an interaction term of sample type and treatment was significant at an ANOVA test (p-value < 0.001) using a model, lmer(species richness ~ sample type + treatment + sample type * treatment + (1|subject_id)). Statistical significances were noted with *: p-value < 0.05, **: p-value < 0.01, and ***: p-value < 0.001.

LMER raw result - Mock

sr_lmer_mock <- lm(S.obs ~ treatment,
                    data = sample_data %>% 
                            data.frame %>% 
                            subset(., .$sample_type %in% c("Mock") & S.obs != 0))

sr_lmer_mock %>% summary() 

## 
## Call:
## lm(formula = S.obs ~ treatment, data = sample_data %>% data.frame %>% 
##     subset(., .$sample_type %in% c("Mock") & S.obs != 0))
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -17.400  -1.100  -0.200   1.067  13.600 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          40.667      2.560  15.883 1.43e-14 ***
## treatmentlyPMA       -5.267      3.798  -1.387 0.177739    
## treatmentBenzonase  -16.467      3.798  -4.336 0.000208 ***
## treatmentHostZERO   -15.667      3.798  -4.125 0.000359 ***
## treatmentMolYsis    -12.267      3.798  -3.230 0.003452 ** 
## treatmentQIAamp     -15.467      3.798  -4.073 0.000411 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 6.272 on 25 degrees of freedom
## Multiple R-squared:  0.5595, Adjusted R-squared:  0.4714 
## F-statistic:  6.35 on 5 and 25 DF,  p-value: 0.0006198

LMER raw result - BAL

sr_lmer_bal <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data %>% 
                            data.frame %>% 
                            subset(., .$sample_type %in% c("BAL") & S.obs != 0))
sr_lmer_bal %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL") &  
##     S.obs != 0)
## 
## REML criterion at convergence: 179
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.45497 -0.50930 -0.00625  0.43924  2.32013 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 90.44    9.510   
##  Residual               94.26    9.709   
## Number of obs: 28, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error     df t value Pr(>|t|)  
## (Intercept)           4.919      6.526 12.892   0.754   0.4646  
## treatmentlyPMA        1.750      6.865 18.131   0.255   0.8017  
## treatmentBenzonase    5.681      6.584 18.317   0.863   0.3994  
## treatmentHostZERO     8.881      6.584 18.317   1.349   0.1938  
## treatmentMolYsis     18.881      6.584 18.317   2.868   0.0101 *
## treatmentQIAamp       9.481      6.584 18.317   1.440   0.1667  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.526                            
## trtmntBnzns -0.570  0.521                     
## trtmntHZERO -0.570  0.521  0.565              
## trtmntMlYss -0.570  0.521  0.565  0.565       
## trtmntQIAmp -0.570  0.521  0.565  0.565  0.565

LMER raw result - Nasal

sr_lmer_ns <- lmer(S.obs ~ treatment + (1|subject_id),
                   data = sample_data %>% 
                           data.frame %>% 
                           subset(., .$sample_type %in% c("Nasal")))
sr_lmer_ns %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal"))
## 
## REML criterion at convergence: 179.7
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.93714 -0.54065 -0.09294  0.62090  3.07984 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept)  1.308   1.144   
##  Residual               18.894   4.347   
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)         10.4000     1.4213 28.7781   7.317 4.87e-08 ***
## treatmentlyPMA      -4.8090     2.4049 25.1935  -2.000 0.056431 .  
## treatmentBenzonase  -0.3632     2.4053 25.2997  -0.151 0.881180    
## treatmentHostZERO   10.0368     2.4053 25.2997   4.173 0.000311 ***
## treatmentMolYsis     6.2090     2.4049 25.1935   2.582 0.016026 *  
## treatmentQIAamp      7.7632     2.4053 25.2997   3.227 0.003440 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.553                            
## trtmntBnzns -0.553  0.314                     
## trtmntHZERO -0.553  0.314  0.347              
## trtmntMlYss -0.553  0.307  0.339  0.339       
## trtmntQIAmp -0.553  0.339  0.306  0.306  0.314

LMER raw result - Sputum

sr_lmer_spt <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data %>% 
                            data.frame %>% 
                            subset(., .$sample_type %in% c("Sputum")))
sr_lmer_spt %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum"))
## 
## REML criterion at convergence: 218.6
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -2.0627 -0.4486 -0.1189  0.3630  1.7435 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 322.9    17.97   
##  Residual               245.7    15.68   
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          15.800     10.664   9.187   1.482  0.17192    
## treatmentlyPMA       37.600      9.914  20.000   3.793  0.00114 ** 
## treatmentBenzonase   66.600      9.914  20.000   6.718 1.55e-06 ***
## treatmentHostZERO   103.000      9.914  20.000  10.390 1.66e-09 ***
## treatmentMolYsis    112.800      9.914  20.000  11.378 3.46e-10 ***
## treatmentQIAamp      85.200      9.914  20.000   8.594 3.79e-08 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.465                            
## trtmntBnzns -0.465  0.500                     
## trtmntHZERO -0.465  0.500  0.500              
## trtmntMlYss -0.465  0.500  0.500  0.500       
## trtmntQIAmp -0.465  0.500  0.500  0.500  0.500

sr_lmer_spt %>% confint()

##                        2.5 %    97.5 %
## .sig01              8.318892  36.73323
## .sigma             10.877929  19.02664
## (Intercept)        -4.977360  36.57737
## treatmentlyPMA     19.531174  55.66883
## treatmentBenzonase 48.531174  84.66883
## treatmentHostZERO  84.931174 121.06883
## treatmentMolYsis   94.731174 130.86883
## treatmentQIAamp    67.131174 103.26883

Tidy summarized table

sr_lmer_mock_kbl <-  sr_lmer_mock %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_mock %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        

sr_lmer_bal_kbl <-  sr_lmer_bal %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
sr_lmer_ns_kbl <-  sr_lmer_ns %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


sr_lmer_spt_kbl <-  sr_lmer_spt %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        

tables3 <- cbind(#sr_lmer_mock_kbl,
                 sr_lmer_bal_kbl,
                 sr_lmer_ns_kbl,
                 sr_lmer_spt_kbl) %>%
    kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 1,
                           #"Mock" = 3,
                           "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")

tables3

save_kable(tables3, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS3.html", self_contained = T)

Table S3 viral. Effect size, standard error (SE) and p-value of a statistical test for viral species richness using linear mixed effect model stratified by sample type (Species richness ~ treatment + (1|subject_id) ). Stratified analyses were conducted for each sample type as an interaction term of sample type and treatment was significant at an ANOVA test (p-value < 0.001) using a model, lmer(species richness ~ sample type + treatment + sample type * treatment + (1|subject_id)). Statistical significances were noted with *: p-value < 0.05, **: p-value < 0.01, and ***: p-value < 0.001.

LMER raw result - BAL

sr_lmer_bal_v <- lmerTest::lmer(V.obs ~ treatment + (1|subject_id),
                    data = sample_data %>% 
                            data.frame %>% 
                           subset(., .$sample_type %in% c("BAL")))
sr_lmer_bal_v %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: V.obs ~ treatment + (1 | subject_id)
##    Data: sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL"))
## 
## REML criterion at convergence: 197.7
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.72110 -0.48044 -0.00556  0.48162  2.20074 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 113.3    10.64   
##  Residual               106.1    10.30   
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error     df t value Pr(>|t|)  
## (Intercept)           0.200      6.624 10.288   0.030   0.9765  
## treatmentlyPMA        0.400      6.515 20.000   0.061   0.9517  
## treatmentBenzonase    5.600      6.515 20.000   0.860   0.4002  
## treatmentHostZERO     6.600      6.515 20.000   1.013   0.3231  
## treatmentMolYsis     15.000      6.515 20.000   2.302   0.0322 *
## treatmentQIAamp      10.600      6.515 20.000   1.627   0.1194  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.492                            
## trtmntBnzns -0.492  0.500                     
## trtmntHZERO -0.492  0.500  0.500              
## trtmntMlYss -0.492  0.500  0.500  0.500       
## trtmntQIAmp -0.492  0.500  0.500  0.500  0.500

LMER raw result - Nasal

sr_lmer_ns_v <- lmerTest::lmer(V.obs ~ treatment + (1|subject_id),
                   data = sample_data %>% 
                           data.frame %>% 
                           subset(., .$sample_type %in% c("Nasal")))
sr_lmer_ns_v %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: V.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal"))
## 
## REML criterion at convergence: 187
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.36349 -0.67373  0.06339  0.55537  1.43318 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 16.61    4.075   
##  Residual               16.70    4.086   
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          9.6000     1.8249 18.4329   5.261 4.91e-05 ***
## treatmentlyPMA      -6.4476     2.3678 22.2608  -2.723   0.0123 *  
## treatmentBenzonase  -3.8718     2.3780 22.5323  -1.628   0.1174    
## treatmentHostZERO    4.1282     2.3780 22.5323   1.736   0.0962 .  
## treatmentMolYsis    -0.1524     2.3678 22.2608  -0.064   0.9492    
## treatmentQIAamp      4.0718     2.3780 22.5323   1.712   0.1006    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.386                            
## trtmntBnzns -0.385  0.216                     
## trtmntHZERO -0.385  0.216  0.409              
## trtmntMlYss -0.386  0.191  0.377  0.377       
## trtmntQIAmp -0.385  0.377  0.181  0.181  0.216

LMER raw result - Sputum

sr_lmer_spt_v <- lmerTest::lmer(V.obs ~ treatment + (1|subject_id),
                    data = sample_data %>% 
                            data.frame %>% 
                            subset(., .$sample_type %in% c("Sputum")))
sr_lmer_spt_v %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: V.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum"))
## 
## REML criterion at convergence: 235.1
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.9580 -0.5365 -0.1189  0.4298  1.8857 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 579.2    24.07   
##  Residual               804.2    28.36   
## Number of obs: 29, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)           9.193     17.977  14.856   0.511 0.616615    
## treatmentlyPMA        4.207     19.188  19.086   0.219 0.828770    
## treatmentBenzonase   19.207     19.188  19.086   1.001 0.329340    
## treatmentHostZERO    91.807     19.188  19.086   4.785 0.000127 ***
## treatmentMolYsis    118.407     19.188  19.086   6.171 6.13e-06 ***
## treatmentQIAamp      46.607     19.188  19.086   2.429 0.025183 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.601                            
## trtmntBnzns -0.601  0.563                     
## trtmntHZERO -0.601  0.563  0.563              
## trtmntMlYss -0.601  0.563  0.563  0.563       
## trtmntQIAmp -0.601  0.563  0.563  0.563  0.563

Tidy summarized table

sr_lmer_bal_kbl_v <-  sr_lmer_bal_v %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
sr_lmer_ns_kbl_v <-  sr_lmer_ns_v %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


sr_lmer_spt_kbl_v <-  sr_lmer_spt_v %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
 
tables3 <- 
        
        rbind(
                cbind(Outcome = "Microbial species richness",
                      sr_lmer_bal_kbl %>% rownames_to_column("Treatment"), 
                      sr_lmer_ns_kbl, 
                      sr_lmer_spt_kbl) %>% remove_rownames() %>% as.matrix(),
                cbind(Outcome = "Viral species richness",
                      sr_lmer_bal_kbl_v %>% rownames_to_column("Treatment"), 
                      sr_lmer_ns_kbl_v, 
                      sr_lmer_spt_kbl_v) %>% remove_rownames() %>% as.matrix()
                ) %>%
    kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 2,
                           #"Mock" = 3,
                           "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")

tables3

save_kable(tables3, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS3_updated.html", self_contained = T)

Species richness - stratified - with depth

This table is too redundant. Removed from the manuscript..

Sequencing depth adjusted effect size, standard error (SE) and p-value of a statistical test for species richness using linear mixed effect model stratified by sample type (Species richness ~ treatment + log10 (Final_reads) + (1|subject_id) ).

sr_lmer_bal_w_depth <- lmer(S.obs ~ treatment + bal_log_centered_final_reads + (1|subject_id), data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL") & S.obs != 0)) %>% 
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        mutate("Effect size (95% CI)" = paste(round(Estimate, 1) %>% format(nsmall = 1), 
                                " (", 
                                round(Estimate - 1.96 * abs(t_value), 1) %>% format(nsmall = 1),
                                ", ",
                                round(Estimate + 1.96 * abs(t_value), 1) %>% format(nsmall = 1),
                                ")", 
                                sep = ""),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " "))
        
sr_lmer_ns_w_depth <- lmer(S.obs ~ treatment + ns_log_centered_final_reads + (1|subject_id), data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal"))) %>% 
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("ns_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        mutate("Effect size (95% CI)" = paste(round(Estimate, 1) %>% format(nsmall = ), 
                                " (", 
                                round(Estimate - 1.96 * abs(t_value), 1) %>% format(nsmall = ),
                                ", ",
                                round(Estimate + 1.96 * abs(t_value), 1) %>% format(nsmall = ),
                                ")", 
                                sep = ""),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " "))


sr_lmer_spt_w_depth <- lmer(S.obs ~ treatment + spt_log_centered_final_reads + (1|subject_id), data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum"))) %>% 
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>%
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("spt_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
                rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        mutate("Effect size (95% CI)" = paste(round(Estimate, 1) %>% format(nsmall = 1), 
                                " (", 
                                round(Estimate - 1.96 * abs(t_value), 1) %>% format(nsmall = 1),
                                ", ",
                                round(Estimate + 1.96 * abs(t_value), 1) %>% format(nsmall = 1),
                                ")", 
                                sep = ""),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " "))
        

cbind(sr_lmer_bal_w_depth, sr_lmer_ns_w_depth, sr_lmer_spt_w_depth) %>%
    kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 1, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")

#save_kable(tableS7, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS7.html", self_contained = T)

PERMANOVA - all samples

This table is too redundant. Removed from the manuscript..

Degree of freedom, effect size (residual, R2) and p-value of permutational ANOVA for Morisita-Horn dissimilarity with an interaction term and strata term (MH-index of species composition ~ sample type * treatment + subject + log10(final reads), strata = subject id). Statistical significances were noted with ***: p-value < 0.001.

set.seed(seed)

phyloseq_rel_nz_neg <- phyloseq$phyloseq_rel %>%
        subset_samples(S.obs != 0 & sample_type %in% c("Neg.", "Mock", 
                                                       "BAL", "Nasal", "Sputum"))

horn_perm_all_neg <- vegan::adonis2(by = "terms",
               distance(phyloseq_rel_nz_neg %>%
                                subset_samples(), method="horn") ~ bal + ns + spt + mock,
               data = phyloseq_rel_nz_neg %>% sample_data %>% data.frame(check.names = F) %>%
                       mutate(neg = ifelse(sample_type == "Neg.", 1, 0),
                              bal = ifelse(sample_type == "BAL", 1, 0),
                              ns = ifelse(sample_type == "Nasal", 1, 0),
                              spt = ifelse(sample_type == "Sputum", 1, 0),
                              mock = ifelse(sample_type == "Mock", 1, 0)),
               permutations = 10000)

set.seed(seed)

phyloseq_rel_nz_neg_decontam <- phyloseq_decontam$phyloseq_rel %>%
        subset_samples(S.obs != 0 & sample_type %in% c("Neg.", "Mock", 
                                                       "BAL", "Nasal", "Sputum"))

horn_perm_all_neg_decontam <- vegan::adonis2(by = "terms",
               distance(phyloseq_rel_nz_neg_decontam %>%
                                subset_samples(), method="horn") ~ bal + ns + spt + mock,
               data = phyloseq_rel_nz_neg_decontam %>% sample_data %>% data.frame(check.names = F) %>%
                       mutate(neg = ifelse(sample_type == "Neg.", 1, 0),
                              bal = ifelse(sample_type == "BAL", 1, 0),
                              ns = ifelse(sample_type == "Nasal", 1, 0),
                              spt = ifelse(sample_type == "Sputum", 1, 0),
                              mock = ifelse(sample_type == "Mock", 1, 0)),
               permutations = 10000)


set.seed(seed)
horn_perm_all <- vegan::adonis2(by = "terms",
                                distance(phyloseq_rel_nz, method="horn") ~ sample_type * treatment + subject_id,
                                  data = phyloseq_rel_nz %>% sample_data %>% data.frame(check.names = F),
                                  permutations = 10000)
set.seed(seed)
horn_perm_ns <- vegan::adonis2(by = "terms", distance(subset_samples(phyloseq_rel_nz, sample_type == "Nasal"), method="horn") ~ lypma + benzonase + host_zero + molysis + qiaamp,
                               data = subset_samples(phyloseq_rel_nz, sample_type == "Nasal") %>%
                                       sample_data %>% data.frame(check.names = F),
                               strata = subset_samples(phyloseq_rel_nz, sample_type == "Nasal") %>% 
                                       sample_data %>% data.frame(check.names = F) %>% .$subject_id, permutations = 10000)
set.seed(seed)
horn_perm_bal  <- vegan::adonis2(by = "terms", distance(subset_samples(phyloseq_rel_nz, sample_type == "BAL"), method="horn") ~  lypma + benzonase + host_zero + molysis + qiaamp,
                                 data = subset_samples(phyloseq_rel_nz, sample_type == "BAL") %>% sample_data %>% data.frame(check.names = F),
                                 strata = subset_samples(phyloseq_rel_nz, sample_type == "BAL") %>%
                                         sample_data %>% data.frame(check.names = F) %>% .$subject_id,
                                  permutations = 10000)
set.seed(seed)
horn_perm_spt <- vegan::adonis2(by = "terms", distance(subset_samples(phyloseq_rel_nz, sample_type == "Sputum"), method="horn") ~ lypma + benzonase + host_zero + molysis + qiaamp,
                                data = subset_samples(phyloseq_rel_nz, sample_type == "Sputum") %>% sample_data %>% data.frame(check.names = F),
                                strata = subset_samples(phyloseq_rel_nz, sample_type == "Sputum")
                                %>% sample_data %>% data.frame(check.names = F) %>% .$subject_id,
                                  permutations = 10000) 

PERMANOVA result

PERM(Morisita-Horn dissimilarity ~ bal + ns + spt + mock), negative controls as reference group.

all the sample types were distinctive.

horn_perm_all_neg

PERMANOVA result after removing potential contaminants (identified from decontam)

PERM(Morisita-Horn dissimilarity ~ bal + ns + spt + mock), negative controls as reference group.

all the sample types were distinctive.

horn_perm_all_neg_decontam

PERMANOVA result without negative control samples

horn_perm_all 

Tidy permanova result

tableS5_A <- horn_perm_all %>% data.frame(check.names = F) %>% rownames_to_column("row.names") %>% 
        mutate(row.names = case_when(row.names == "sample_type" ~ 'Sample type',
                                     row.names == "treatment" ~ 'Treatment',
                                     row.names == "subject_id" ~ 'Subject',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "sample_type:treatment" ~ 'Sample type * Treatment',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
        mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        mutate(`<i>p</i>-value` = format(`<i>p</i>-value`, nsmall = 3)) %>%
        dplyr::select(c("Degree of freedom", "R<sup>2</sup>", "<i>p</i>-value", " ")) %>% 
        kbl(format = "html", escape = 0) %>%
        kable_styling(full_width = 0, html_font = "sans")

tableS5_A

Table S6. PERMANOVA - Stratified

Raw result - BAL

horn_perm_bal

Raw result - Nasal

horn_perm_ns

Raw result - BAL

horn_perm_spt

This table was additionally added to the manuscript as to determine bias after treatment with sputum.

Table S6. Degree of freedom, effect size (residual, R^2) and p-value of permutational ANOVA for Morisita-Horn distiances for species richness (MH-distance ~ lyPMA + Benzoase + HostZERO + MolYsis + QIAamp, strata = subject_id). Sample types were stratified as interaction term of (MH-distance ~ sample type + treatment + sample type * treatment, strata = subject_id) was significant (p < 0.001).

Tidy permanova result (stratified)

a <- horn_perm_bal %>% data.frame(check.names = F) %>% rownames_to_column('row.names') %>% 
        mutate(row.names = case_when(row.names == "lypma" ~ 'lyPMA',
                                     row.names == "benzonase" ~ 'Benzonase',
                                     row.names == "host_zero" ~ 'HostZERO',
                                     row.names == "molysis" ~ 'MolYsis',
                                     row.names == "qiaamp" ~ 'QIAamp',
                                     row.names == "subject_id" ~ 'Subject id',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
                mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        dplyr::select(c("R<sup>2</sup>", "<i>p</i>-value", " ")) 

b <- horn_perm_ns %>% data.frame(check.names = F) %>% rownames_to_column('row.names') %>% 
        mutate(row.names = case_when(row.names == "lypma" ~ 'lyPMA',
                                     row.names == "benzonase" ~ 'Benzonase',
                                     row.names == "host_zero" ~ 'HostZERO',
                                     row.names == "molysis" ~ 'MolYsis',
                                     row.names == "qiaamp" ~ 'QIAamp',
                                     row.names == "subject_id" ~ 'Subject id',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
                mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        dplyr::select(c("R<sup>2</sup>", "<i>p</i>-value", " ")) 

c <- horn_perm_spt %>% data.frame(check.names = F) %>% rownames_to_column('row.names') %>% 
        mutate(row.names = case_when(row.names == "lypma" ~ 'lyPMA',
                                     row.names == "benzonase" ~ 'Benzonase',
                                     row.names == "host_zero" ~ 'HostZERO',
                                     row.names == "molysis" ~ 'MolYsis',
                                     row.names == "qiaamp" ~ 'QIAamp',
                                     row.names == "subject_id" ~ 'Subject id',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
                mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        dplyr::select(c("R<sup>2</sup>", "<i>p</i>-value", " ")) 


tableS6 <- cbind(a, b, c) %>% 
        kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 1, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")


save_kable(tableS6, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS6.html", self_contained = T)


tableS6

Table S7. LM on M-H distance

Table S7. Effect size, standard error (SE) and p-value of a statistical test for Morisita-Horn dissimilarity from untreated to each treated within subject, stratified by sample type. Stratified analyses were conducted for each sample type as an interaction term of sample type and treatment was significant (p-value < 0.001) using a model, ANOVA(species richness ~ sample type + treatment + sample type * treatment). The tested was conducted with a linear mixed effect model LM(distance within subject by treatment ~ treatment + (1|subject)). The baseline of sample type is untreated (0 distance, untreated from untreated), and statistical significance were noted with *: p-value < 0.05, **: p-value < 0.01 and ***:p-value < 0.001.

horn_dist_long_within_sampleid_from_control$treatment <- factor(horn_dist_long_within_sampleid_from_control$treatment,
                                                                     levels = c("Untreated",
                                                                                "lypma",
                                                                                "benzonase",
                                                                                "host_zero",
                                                                                "molysis",
                                                                                "qiaamp"))
horn_dist_long_within_sampleid_from_control$sample_type <- factor(horn_dist_long_within_sampleid_from_control$sample_type,
                                                                     levels = c("BAL",
                                                                                "Nasal",
                                                                                "Sputum"))

Raw lmer result on M-H distances

lmer(dist ~ treatment * sample_type + (1|subject_id),
     data = horn_dist_long_within_sampleid_from_control %>%
             subset(sample_type != "Mock")) %>%
        summary() 

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: dist ~ treatment * sample_type + (1 | subject_id)
##    Data: horn_dist_long_within_sampleid_from_control %>% subset(sample_type !=  
##     "Mock")
## 
## REML criterion at convergence: -4.5
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -2.20682 -0.58433 -0.06776  0.47243  2.61620 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.02288  0.1512  
##  Residual               0.02599  0.1612  
## Number of obs: 89, groups:  subject_id, 19
## 
## Fixed effects:
##                                        Estimate Std. Error         df t value
## (Intercept)                           8.581e-16  1.105e-01  4.166e+01   0.000
## treatmentlypma                        3.252e-01  1.140e-01  5.672e+01   2.852
## treatmentbenzonase                    1.067e-01  1.140e-01  5.672e+01   0.936
## treatmenthost_zero                    2.736e-01  1.140e-01  5.672e+01   2.400
## treatmentmolysis                      2.039e-01  1.140e-01  5.672e+01   1.789
## treatmentqiaamp                       2.508e-01  1.140e-01  5.672e+01   2.200
## sample_typeNasal                     -8.635e-16  1.308e-01  4.166e+01   0.000
## sample_typeSputum                    -1.056e-15  1.483e-01  4.166e+01   0.000
## treatmentlypma:sample_typeNasal      -1.455e-01  1.472e-01  5.885e+01  -0.988
## treatmentbenzonase:sample_typeNasal   3.519e-02  1.475e-01  5.912e+01   0.239
## treatmenthost_zero:sample_typeNasal  -2.428e-01  1.475e-01  5.912e+01  -1.646
## treatmentmolysis:sample_typeNasal     1.194e-02  1.472e-01  5.885e+01   0.081
## treatmentqiaamp:sample_typeNasal     -9.850e-02  1.475e-01  5.912e+01  -0.668
## treatmentlypma:sample_typeSputum      2.215e-02  1.529e-01  5.672e+01   0.145
## treatmentbenzonase:sample_typeSputum  4.126e-01  1.529e-01  5.672e+01   2.697
## treatmenthost_zero:sample_typeSputum  3.383e-01  1.529e-01  5.672e+01   2.212
## treatmentmolysis:sample_typeSputum    3.987e-01  1.529e-01  5.672e+01   2.607
## treatmentqiaamp:sample_typeSputum     3.638e-01  1.529e-01  5.672e+01   2.378
##                                      Pr(>|t|)   
## (Intercept)                           1.00000   
## treatmentlypma                        0.00605 **
## treatmentbenzonase                    0.35332   
## treatmenthost_zero                    0.01972 * 
## treatmentmolysis                      0.07904 . 
## treatmentqiaamp                       0.03187 * 
## sample_typeNasal                      1.00000   
## sample_typeSputum                     1.00000   
## treatmentlypma:sample_typeNasal       0.32697   
## treatmentbenzonase:sample_typeNasal   0.81223   
## treatmenthost_zero:sample_typeNasal   0.10498   
## treatmentmolysis:sample_typeNasal     0.93565   
## treatmentqiaamp:sample_typeNasal      0.50682   
## treatmentlypma:sample_typeSputum      0.88538   
## treatmentbenzonase:sample_typeSputum  0.00919 **
## treatmenthost_zero:sample_typeSputum  0.03102 * 
## treatmentmolysis:sample_typeSputum    0.01166 * 
## treatmentqiaamp:sample_typeSputum     0.02078 * 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Justifying stratified analysis - ANOVA(LM(dist ~ sample type * treatment))

lmer(dist ~ treatment * sample_type + (1|subject_id),
     data = horn_dist_long_within_sampleid_from_control %>%
             subset(sample_type != "Mock")) %>%
        anova() 

Stratified analysis - BAL

mh_lmer_bal <- lmer(dist ~ treatment + (1|subject_id),
     data = horn_dist_long_within_sampleid_from_control %>%
             subset(sample_type == "BAL")) 

mh_lmer_bal %>% 
        summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: dist ~ treatment + (1 | subject_id)
##    Data: horn_dist_long_within_sampleid_from_control %>% subset(sample_type ==  
##     "BAL")
## 
## REML criterion at convergence: 3.8
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.8036 -0.4949 -0.1508  0.5054  2.2725 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.01796  0.1340  
##  Residual               0.03615  0.1901  
## Number of obs: 24, groups:  subject_id, 4
## 
## Fixed effects:
##                     Estimate Std. Error        df t value Pr(>|t|)  
## (Intercept)        2.261e-16  1.163e-01 1.161e+01   0.000   1.0000  
## treatmentlypma     3.252e-01  1.344e-01 1.500e+01   2.419   0.0288 *
## treatmentbenzonase 1.067e-01  1.344e-01 1.500e+01   0.794   0.4398  
## treatmenthost_zero 2.736e-01  1.344e-01 1.500e+01   2.035   0.0599 .
## treatmentmolysis   2.039e-01  1.344e-01 1.500e+01   1.517   0.1502  
## treatmentqiaamp    2.508e-01  1.344e-01 1.500e+01   1.866   0.0818 .
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtmntl trtmntb trtmn_ trtmntm
## tretmntlypm -0.578                               
## trtmntbnzns -0.578  0.500                        
## trtmnthst_z -0.578  0.500   0.500                
## trtmntmlyss -0.578  0.500   0.500   0.500        
## treatmntqmp -0.578  0.500   0.500   0.500  0.500

Stratified analysis - Nasal

mh_lmer_ns <- lmer(dist ~ treatment + (1|subject_id),
     data = horn_dist_long_within_sampleid_from_control %>%
             subset(sample_type == "Nasal")) 

mh_lmer_ns %>% 
        summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: dist ~ treatment + (1 | subject_id)
##    Data: horn_dist_long_within_sampleid_from_control %>% subset(sample_type ==  
##     "Nasal")
## 
## REML criterion at convergence: -28.4
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.4167 -0.6315 -0.1315  0.2369  2.1381 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.002114 0.04598 
##  Residual               0.013564 0.11646 
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                     Estimate Std. Error        df t value Pr(>|t|)   
## (Intercept)        7.015e-17  3.960e-02 2.775e+01   0.000  1.00000   
## treatmentlypma     1.751e-01  6.506e-02 2.378e+01   2.692  0.01279 * 
## treatmentbenzonase 1.600e-01  6.510e-02 2.400e+01   2.457  0.02160 * 
## treatmenthost_zero 4.886e-02  6.510e-02 2.400e+01   0.750  0.46027   
## treatmentmolysis   2.203e-01  6.506e-02 2.378e+01   3.386  0.00246 **
## treatmentqiaamp    1.342e-01  6.510e-02 2.400e+01   2.062  0.05018 . 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtmntl trtmntb trtmn_ trtmntm
## tretmntlypm -0.527                               
## trtmntbnzns -0.526  0.295                        
## trtmnthst_z -0.526  0.295   0.360                
## trtmntmlyss -0.527  0.282   0.346   0.346        
## treatmntqmp -0.526  0.346   0.280   0.280  0.295

Stratified analysis - Sputum

mh_lmer_spt <- lmer(dist ~ treatment + (1|subject_id),
     data = horn_dist_long_within_sampleid_from_control %>%
             subset(sample_type == "Sputum")) 

mh_lmer_spt %>% 
        summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: dist ~ treatment + (1 | subject_id)
##    Data: horn_dist_long_within_sampleid_from_control %>% subset(sample_type ==  
##     "Sputum")
## 
## REML criterion at convergence: 5.6
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.8788 -0.5690  0.0840  0.4564  1.8823 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.06615  0.2572  
##  Residual               0.03208  0.1791  
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                     Estimate Std. Error        df t value Pr(>|t|)    
## (Intercept)        1.186e-15  1.402e-01 7.345e+00   0.000  1.00000    
## treatmentlypma     3.473e-01  1.133e-01 2.000e+01   3.066  0.00610 ** 
## treatmentbenzonase 5.193e-01  1.133e-01 2.000e+01   4.584  0.00018 ***
## treatmenthost_zero 6.119e-01  1.133e-01 2.000e+01   5.402 2.75e-05 ***
## treatmentmolysis   6.026e-01  1.133e-01 2.000e+01   5.320 3.31e-05 ***
## treatmentqiaamp    6.146e-01  1.133e-01 2.000e+01   5.426 2.60e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtmntl trtmntb trtmn_ trtmntm
## tretmntlypm -0.404                               
## trtmntbnzns -0.404  0.500                        
## trtmnthst_z -0.404  0.500   0.500                
## trtmntmlyss -0.404  0.500   0.500   0.500        
## treatmntqmp -0.404  0.500   0.500   0.500  0.500

Viral beta diversity (LMER)

#Making subset of non-zero samples without neg

phyloseq_rel_nz_v <- v_phyloseq$viral_rel %>% 
        subset_samples(S.obs != 0 & sample_type %in% c("Mock", 
                                                       "BAL", "Nasal", "Sputum"))

#distances of betadiversity - boxplots
horn_dist_long <- distance(phyloseq_rel_nz_v, method="horn") %>% as.matrix() %>% melt_dist() #making long data of distance matrices

#Adding sample type and treatment name. 
#this can be also done by merging metadata into the `horn_dist_long`
names <- data.frame(str_split_fixed(horn_dist_long$iso1, "_", 3))
names2 <- data.frame(str_split_fixed(horn_dist_long$iso2, "_", 3))
horn_dist_long$sample_id_1 <- paste(names$X1, names$X2, sep = "_")
horn_dist_long$method_1 <- ifelse(grepl("lyPMA", horn_dist_long$iso1),"lypma", 
                                         ifelse(grepl("ben", horn_dist_long$iso1),"benzonase", 
                                                ifelse(grepl("host", horn_dist_long$iso1),"host_zero", 
                                                       ifelse(grepl("qia", horn_dist_long$iso1),"qiaamp", 
                                                              ifelse(grepl("moly", horn_dist_long$iso1),"molysis", 
                                                                     "control")))))


#Adding data for iso 2 also should be done
horn_dist_long$sample_id_2 <- paste(names2$X1, names2$X2, sep = "_")
horn_dist_long$method_2 <-ifelse(grepl("lyPMA", horn_dist_long$iso2),"lypma", 
                                        ifelse(grepl("ben", horn_dist_long$iso2),"benzonase", 
                                               ifelse(grepl("host", horn_dist_long$iso2),"host_zero", 
                                                      ifelse(grepl("qia", horn_dist_long$iso2),"qiaamp", 
                                                             ifelse(grepl("moly", horn_dist_long$iso2),"molysis", 
                                                                    "control")))))


#subsetting distances of my interest
horn_dist_long$sample_id_1 <- ifelse(grepl("pos", horn_dist_long$sample_id_1, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", horn_dist_long$sample_id_1, ignore.case = T),"Neg.",
                                        horn_dist_long$sample_id_1))
horn_dist_long$sample_id_2 <- ifelse(grepl("pos", horn_dist_long$sample_id_2, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", horn_dist_long$sample_id_2, ignore.case = T),"Neg.",
                                        horn_dist_long$sample_id_2))


horn_dist_long_within_sampleid_from_control <- subset(horn_dist_long, horn_dist_long$sample_id_1 == horn_dist_long$sample_id_2) # data within samples

horn_dist_long_within_sampleid_from_control <- subset(horn_dist_long_within_sampleid_from_control,
                                                           horn_dist_long_within_sampleid_from_control$method_1 != horn_dist_long_within_sampleid_from_control$method_2) # remove irrelevant association

horn_dist_long_within_sampleid_from_control <- subset(horn_dist_long_within_sampleid_from_control, (horn_dist_long_within_sampleid_from_control$method_1 == "control") + (horn_dist_long_within_sampleid_from_control$method_2 == "control") != 0)


horn_dist_long_within_sampleid_from_control$treatment <- horn_dist_long_within_sampleid_from_control$method_1

horn_dist_long_within_sampleid_from_control$treatment <- ifelse(horn_dist_long_within_sampleid_from_control$treatment == "control", horn_dist_long_within_sampleid_from_control$method_2, horn_dist_long_within_sampleid_from_control$treatment) 


#Setting key method
horn_dist_long_within_sampleid_from_control$sample_type <- ifelse(grepl("NS", horn_dist_long_within_sampleid_from_control$iso1), "Nasal",
                                                                  ifelse(grepl("CFB", horn_dist_long_within_sampleid_from_control$iso1), "Sputum",
                                                                         ifelse(grepl("BAL", horn_dist_long_within_sampleid_from_control$iso1), "BAL",
                                                                                ifelse(grepl("pos|POS", horn_dist_long_within_sampleid_from_control$iso1, ignore.case = T), "Mock",
                                                                                       ifelse(grepl("neg|N_EXT", horn_dist_long_within_sampleid_from_control$iso1), "Neg.",NA)))))

#Making a column for baseline (controls, from where?)
horn_dist_long_within_sampleid_from_control <- horn_dist_long_within_sampleid_from_control %>% 
        mutate(dist_from = case_when(method_1 == "control" ~ iso1,
                                     method_2 == "control" ~ iso2))

dummy <- data.frame(iso1 = horn_dist_long_within_sampleid_from_control$dist_from %>% unique,
           iso2 = horn_dist_long_within_sampleid_from_control$dist_from %>% unique,
           dist = 0,
           treatment = "Untreated",
           method_1 = "control",
           method_2 = "control"
           )
names <- data.frame(str_split_fixed(dummy$iso1, "_", 3))
names2 <- data.frame(str_split_fixed(dummy$iso2, "_", 3))
dummy$sample_id_1 <- paste(names$X1, names$X2, sep = "_")
#Adding data for iso 2 also should be done
dummy$sample_id_2 <- paste(names2$X1, names2$X2, sep = "_")


#subsetting distances of my interest
dummy$sample_id_1 <- ifelse(grepl("pos", dummy$sample_id_1, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", dummy$sample_id_1, ignore.case = T),"Neg.",
                                        dummy$sample_id_1))
dummy$sample_id_2 <- ifelse(grepl("pos", dummy$sample_id_2, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", dummy$sample_id_2, ignore.case = T),"Neg.",
                                        dummy$sample_id_2))
dummy$sample_type <- ifelse(grepl("NS", dummy$iso1), "Nasal",
                            ifelse(grepl("CFB", dummy$iso1), "Sputum",
                                   ifelse(grepl("BAL", dummy$iso1), "BAL",
                                          ifelse(grepl("pos|POS", dummy$iso1, ignore.case = T), "Mock",
                                                 ifelse(grepl("neg|N_EXT", dummy$iso1), "Neg.",NA)))))
dummy <- subset(dummy, !is.na(dummy$sample_type))
horn_dist_long_within_sampleid_from_control <- bind_rows(horn_dist_long_within_sampleid_from_control, dummy)

#Here, sample id is the same as subject id.
horn_dist_long_within_sampleid_from_control$subject_id <- horn_dist_long_within_sampleid_from_control$sample_id_1

horn_dist_long_within_sampleid_from_control$treatment <-
        factor(horn_dist_long_within_sampleid_from_control$treatment,
               levels = c("Untreated", "lypma", "benzonase", "host_zero", "molysis", "qiaamp"))


#mh_lmer_bal_v <- lmer(dist ~ treatment + (1|subject_id),
#     data = horn_dist_long_within_sampleid_from_control %>%
#             subset(sample_type == "BAL")) 

mh_lmer_ns_v <- lmer(dist ~ treatment + (1|subject_id),
     data = horn_dist_long_within_sampleid_from_control %>%
             subset(sample_type == "Nasal")) 

mh_lmer_spt_v <- lmer(dist ~ treatment + (1|subject_id),
     data = horn_dist_long_within_sampleid_from_control %>%
             subset(sample_type == "Sputum")) 

Tidy, summarized-stratified analysis

mh_lmer_bal_kbl <-  mh_lmer_bal %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              mh_lmer_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column("treatment") %>%
        mutate(treatment = case_when(treatment == "(Intercept)" ~ "Untreated",
                                             treatment == "treatmentlypma" ~ "lyPMA",
                                             treatment == "treatmentbenzonase" ~ "Benzonase",
                                             treatment == "treatmenthost_zero" ~ "HostZERO",
                                             treatment == "treatmentmolysis" ~ "MolYsis",
                                             treatment == "treatmentqiaamp" ~ "QIAamp"
                                             )) %>%
        column_to_rownames("treatment") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("Untreated",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


mh_lmer_ns_kbl <-  mh_lmer_ns %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              mh_lmer_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                column_to_rownames("Row.names") %>%
        rownames_to_column("treatment") %>%
        mutate(treatment = case_when(treatment == "(Intercept)" ~ "Untreated",
                                             treatment == "treatmentlypma" ~ "lyPMA",
                                             treatment == "treatmentbenzonase" ~ "Benzonase",
                                             treatment == "treatmenthost_zero" ~ "HostZERO",
                                             treatment == "treatmentmolysis" ~ "MolYsis",
                                             treatment == "treatmentqiaamp" ~ "QIAamp"
                                             )) %>%
        column_to_rownames("treatment")  %>% 
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("Untreated",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


mh_lmer_spt_kbl <-  mh_lmer_spt %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              mh_lmer_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                column_to_rownames("Row.names") %>%
        rownames_to_column("treatment") %>%
        mutate(treatment = case_when(treatment == "(Intercept)" ~ "Untreated",
                                             treatment == "treatmentlypma" ~ "lyPMA",
                                             treatment == "treatmentbenzonase" ~ "Benzonase",
                                             treatment == "treatmenthost_zero" ~ "HostZERO",
                                             treatment == "treatmentmolysis" ~ "MolYsis",
                                             treatment == "treatmentqiaamp" ~ "QIAamp"
                                             )) %>%
        column_to_rownames("treatment") %>% 
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("Untreated",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]



mh_lmer_bal_kbl_v <-  mh_lmer_bal %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              mh_lmer_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column("treatment") %>%
        mutate(treatment = case_when(treatment == "(Intercept)" ~ "Untreated",
                                             treatment == "treatmentlypma" ~ "lyPMA",
                                             treatment == "treatmentbenzonase" ~ "Benzonase",
                                             treatment == "treatmenthost_zero" ~ "HostZERO",
                                             treatment == "treatmentmolysis" ~ "MolYsis",
                                             treatment == "treatmentqiaamp" ~ "QIAamp"
                                             )) %>%
        column_to_rownames("treatment") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = "-",
               "<i>p</i>-value" = "-",
               " " = "") %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("Untreated",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


mh_lmer_ns_kbl_v <-  mh_lmer_ns_v %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              mh_lmer_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                column_to_rownames("Row.names") %>%
        rownames_to_column("treatment") %>%
        mutate(treatment = case_when(treatment == "(Intercept)" ~ "Untreated",
                                             treatment == "treatmentlypma" ~ "lyPMA",
                                             treatment == "treatmentbenzonase" ~ "Benzonase",
                                             treatment == "treatmenthost_zero" ~ "HostZERO",
                                             treatment == "treatmentmolysis" ~ "MolYsis",
                                             treatment == "treatmentqiaamp" ~ "QIAamp"
                                             )) %>%
        column_to_rownames("treatment")  %>% 
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("Untreated",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


mh_lmer_spt_kbl_v <-  mh_lmer_spt_v %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              mh_lmer_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                column_to_rownames("Row.names") %>%
        rownames_to_column("treatment") %>%
        mutate(treatment = case_when(treatment == "(Intercept)" ~ "Untreated",
                                             treatment == "treatmentlypma" ~ "lyPMA",
                                             treatment == "treatmentbenzonase" ~ "Benzonase",
                                             treatment == "treatmenthost_zero" ~ "HostZERO",
                                             treatment == "treatmentmolysis" ~ "MolYsis",
                                             treatment == "treatmentqiaamp" ~ "QIAamp"
                                             )) %>%
        column_to_rownames("treatment") %>% 
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("Untreated",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]



tableS7 <- 
        rbind(
                cbind(Outcome = "Microbial beta diversity",
                      mh_lmer_bal_kbl %>% rownames_to_column("Treatment"), 
                      mh_lmer_ns_kbl, 
                      mh_lmer_spt_kbl) %>% remove_rownames() %>% as.matrix(),
                cbind(Outcome = "Viral beta diversity",
                      mh_lmer_bal_kbl_v %>% rownames_to_column("Treatment"), 
                      mh_lmer_ns_kbl_v, 
                      mh_lmer_spt_kbl_v) %>% remove_rownames() %>% as.matrix()
                ) %>%
        
        kbl(format = "html", escape = 0) %>% 
        add_header_above(c(" " = 2, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        #add_rownames(c("Microbial species" = 6, "Viral species"= 6)) %>% 
        kable_styling(full_width = 0, html_font = "sans")


save_kable(tableS7, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS7.html", self_contained = T)

tableS7

Beta diversity plot on all samples

fig_sample_type <- ordinate(phyloseq_rel_nz_neg, method = "PCoA", distance = "horn") %>%
        plot_ordination(phyloseq_rel_nz_neg, ., col = "sample_type") +
        #scale_color_viridis(discrete = 6, name = "Treatment", labels = c("Mock theoretical", "Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        scale_color_manual(values = c("#a8ddb5", "#43a2ca", "#a50f15", "#fc9272", "#fee0d2"),
                           name = "Sample type",
                           breaks = c("Neg.","Mock", "BAL", "Nasal", "Sputum"),
                           labels = c("Neg.","Mock", "BAL", "Nasal", "Sputum")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        #scale_shape(name = "Sample type", labels = c("Mock theoretical", "Mock")) +
        geom_point(size = 3) +
        theme_classic (base_size = 12, base_family = "sans") +
        #facet_wrap(~sample_type, scales = "free") +
        #labs(tag = "A") +
        ggtitle("PCoA plot, M-H dissimilarity, on all samples") +
        theme(plot.tag = element_text(size = 15), legend.position = "top") #+
        #stat_ellipse(type = "norm", linetype = 2, linewidth = 0.1) +
        #stat_ellipse(type = "t", linewidth = 0.1)

fig_sample_type

Beta diversity plot of decontaminated samples

fig_sample_type_decontam <- ordinate(phyloseq_rel_nz_neg_decontam, method = "PCoA", distance = "horn") %>%
        plot_ordination(phyloseq_rel_nz_neg_decontam, ., col = "sample_type") +
        #scale_color_viridis(discrete = 6, name = "Treatment", labels = c("Mock theoretical", "Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        scale_color_manual(values = c("#a8ddb5", "#43a2ca", "#a50f15", "#fc9272", "#fee0d2"),
                           name = "Sample type",
                           breaks = c("Neg.","Mock", "BAL", "Nasal", "Sputum"),
                           labels = c("Neg.","Mock", "BAL", "Nasal", "Sputum")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        #scale_shape(name = "Sample type", labels = c("Mock theoretical", "Mock")) +
        geom_point(size = 3) +
        theme_classic (base_size = 12, base_family = "sans") +
        #facet_wrap(~sample_type, scales = "free") +
        #labs(tag = "A") +
        ggtitle("PCoA plot, M-H dissimilarity, on all samples (decontaminated)") +
        theme(plot.tag = element_text(size = 15), legend.position = "top") #+
        #stat_ellipse(type = "norm", linetype = 2, linewidth = 0.1) +
        #stat_ellipse(type = "t", linewidth = 0.1)

fig_sample_type_decontam

Fig. S7 PCA figure

Fig. S7. Principal coordinate analysis plot based on Morisita-Horn dissimilarity of taxonomic sequencing results stratified by sample type.

figS7 <- ordinate(subset_samples(phyloseq_rel_nz, sample_type != "Neg." & sample_type != "Mock"), method = "PCoA", distance = "horn") %>%
        plot_ordination(phyloseq_rel_nz, ., col = "treatment") +
        #scale_color_viridis(discrete = 6, name = "Treatment", labels = c("Mock theoretical", "Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                           name = "Treatment",
                           breaks = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
                           labels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        #scale_shape(name = "Sample type", labels = c("Mock theoretical", "Mock")) +
        geom_point(size = 3) +
        theme_classic (base_size = 12, base_family = "sans") +
        facet_wrap(~sample_type, scales = "free") +
        theme(plot.tag = element_text(size = 15), legend.position = "top")# +
        #stat_ellipse(type = "norm") +
        #stat_ellipse(type = "t")

figS7

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS7.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 90, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue
figS7
dev.off()

## quartz_off_screen 
##                 2

Table S4. Mediation analysis (treatment-stratified)

outcome = S.obs exposure = treatment (stratified) mediator = Final_reads mediator-outcome confounders = sample_type exposure-mediator confounders = NA outcome model = Mixed effects linear regression mediator model = Mixed effects linear regression

Mediation analysis was conducted stratified treatment.

Table S4. Mediation analysis results in estimated effect sizes and p-values of indirect effect, direct effect, and proportion of mediation. The analysis employed treatment as exposure, log10(final reads) as mediator, species richness as outcome, and sample type as mediator-exposure confounder. Analysis was stratified by each treatment and treated as binary variables.

## lypma
# only mediator-outcome confounders
detach_package <- function(pkg, character.only = FALSE)
{
  if(!character.only)
  {
    pkg <- deparse(substitute(pkg))
  }
  search_item <- paste("package", pkg, sep = ":")
  while(search_item %in% search())
  {
    detach(search_item, unload = TRUE, character.only = TRUE)
  }
}

detach_package(lmerTest)

## all treatment groups

sample_data_respiratory <- subset(phyloseq_rel_nz %>% sample_data %>% data.frame, sample_data$sample_type == "Sputum" | sample_data$sample_type == "BAL" | sample_data$sample_type == "Nasal")


med.fit <- lmer(log10(Final_reads) ~ lypma +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$lypma == 1))

out.fit <- lmer(S.obs ~ lypma * log10(Final_reads) + sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$lypma == 1))
set.seed(seed)
med.out.lypma <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "lypma",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.lypma)

final.out.lypma <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## benzonase


med.fit <- lmer(log10(Final_reads) ~ benzonase + sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$benzonase == 1))

out.fit <- lmer(S.obs ~ benzonase * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$benzonase == 1))

set.seed(seed)
med.out.benzonase <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "benzonase",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.benzonase)

final.out.benzonase <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## host zero


med.fit <- lmer(log10(Final_reads) ~ host_zero +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$host_zero == 1))

out.fit <- lmer(S.obs ~ host_zero * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$host_zero == 1))
set.seed(seed)
med.out.hostzero <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "host_zero",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.hostzero)

final.out.host_zero <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))


## molysis


med.fit <- lmer(log10(Final_reads) ~ molysis +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$molysis == 1))

out.fit <- lmer(S.obs ~ molysis * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$molysis == 1))
set.seed(seed)
med.out.molysis <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "molysis",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.molysis)

final.out.molysis <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## qiaamp


med.fit <- lmer(log10(Final_reads) ~ qiaamp +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$qiaamp == 1))

out.fit <- lmer(S.obs ~ qiaamp * log10(Final_reads) + sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$qiaamp == 1))

set.seed(seed)
med.out_qiaamp <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "qiaamp",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out_qiaamp)

final.out.qiaamp <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

Raw mediation outputs - lyPMA

med.out.lypma %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value  
## ACME (control)             0.5058      -3.1990         4.95   0.781  
## ACME (treated)             1.6851      -9.3365        13.54   0.759  
## ADE (control)             11.3959       0.3340        22.95   0.043 *
## ADE (treated)             12.5751       1.2206        24.90   0.028 *
## Total Effect              13.0810       0.3890        26.65   0.043 *
## Prop. Mediated (control)   0.0214      -0.7431         0.46   0.756  
## Prop. Mediated (treated)   0.1321      -2.3171         1.17   0.729  
## ACME (average)             1.0955      -5.9701         8.64   0.759  
## ADE (average)             11.9855       1.2858        22.91   0.027 *
## Prop. Mediated (average)   0.0768      -1.5381         0.77   0.729  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Raw mediation outputs - Benzonase

med.out.benzonase %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value   
## ACME (control)             8.3239      -2.1981        21.41  0.1332   
## ACME (treated)            14.6853      -0.6295        34.13  0.0630 . 
## ADE (control)              8.7025     -11.0395        27.78  0.3824   
## ADE (treated)             15.0638      -3.8177        34.70  0.1208   
## Total Effect              23.3877       7.6197        40.06  0.0038 **
## Prop. Mediated (control)   0.3463      -0.1061         1.32  0.1370   
## Prop. Mediated (treated)   0.6117      -0.0421         1.91  0.0668 . 
## ACME (average)            11.5046       1.3646        24.40  0.0224 * 
## ADE (average)             11.8832      -5.1716        29.09  0.1738   
## Prop. Mediated (average)   0.4790       0.0584         1.46  0.0262 * 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 35 
## 
## 
## Simulations: 10000

Raw mediation outputs - host zero

med.out.hostzero %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)             24.110       -1.084        52.51  0.0626 .  
## ACME (treated)             34.365       10.979        62.01  0.0030 ** 
## ADE (control)               4.994      -22.338        31.79  0.7176    
## ADE (treated)              15.249      -16.203        46.36  0.3390    
## Total Effect               39.359       20.836        58.64  0.0002 ***
## Prop. Mediated (control)    0.608       -0.028         1.57  0.0628 .  
## Prop. Mediated (treated)    0.868        0.277         1.79  0.0032 ** 
## ACME (average)             29.238        9.775        51.95  0.0014 ** 
## ADE (average)              10.121      -14.700        35.03  0.4274    
## Prop. Mediated (average)    0.738        0.243         1.57  0.0016 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 35 
## 
## 
## Simulations: 10000

Raw mediation outputs - MolYsis

med.out.molysis %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)            16.6842      -1.6638        39.14   0.077 .  
## ACME (treated)            36.1230      17.3570        59.33  <2e-16 ***
## ADE (control)              8.9818     -11.1781        28.19   0.364    
## ADE (treated)             28.4205       4.3635        55.21   0.022 *  
## Total Effect              45.1047      26.3931        64.04  <2e-16 ***
## Prop. Mediated (control)   0.3659      -0.0373         0.88   0.077 .  
## Prop. Mediated (treated)   0.7951       0.4294         1.31  <2e-16 ***
## ACME (average)            26.4036      11.8607        44.49  <2e-16 ***
## ADE (average)             18.7011       0.4694        37.76   0.041 *  
## Prop. Mediated (average)   0.5805       0.2833         0.99  <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 35 
## 
## 
## Simulations: 10000

Raw mediation outputs - qiaamp

med.out_qiaamp %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)            22.3244      -2.1803        49.12   0.077 .  
## ACME (treated)            20.2975      -4.2164        46.66   0.111    
## ADE (control)             12.0883     -15.5013        39.45   0.398    
## ADE (treated)             10.0614     -19.4275        38.73   0.506    
## Total Effect              32.3858      17.1994        48.04  <2e-16 ***
## Prop. Mediated (control)   0.6853      -0.0724         1.85   0.077 .  
## Prop. Mediated (treated)   0.6266      -0.1365         1.64   0.111    
## ACME (average)            21.3110       0.8868        43.47   0.039 *  
## ADE (average)             11.0748     -13.2950        35.73   0.383    
## Prop. Mediated (average)   0.6559       0.0266         1.61   0.039 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 35 
## 
## 
## Simulations: 10000

Tidy table of mediation analysis result

tableS4_a <- rbind(final.out.lypma,
      final.out.benzonase,
      final.out.host_zero, 
      final.out.molysis,
      final.out.qiaamp) %>%
mutate(Treatment = c("lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"), .before = "Indirect est.") 

tableS4_a

Mediation analysis for function

outcome = S.obs (with phyloseq$) exposure = treatment (stratified) mediator = Final_reads mediator-outcome confounders = sample_type exposure-mediator confounders = NA outcome model = Mixed effects linear regression mediator model = Mixed effects linear regression

Mediation analysis was conducted stratified treatment.

Table S4. Mediation analysis results in estimated effect sizes and p-values of indirect effect, direct effect, and proportion of mediation. The analysis employed treatment as exposure, log10(final reads) as mediator, species richness as outcome, and sample type as mediator-exposure confounder. Analysis was stratified by each treatment and treated as binary variables.

detach_package(lmerTest)

## all treatment groups

sample_data_respiratory_path <- subset(phyloseq$phyloseq_path_rpk %>% sample_data %>% data.frame, sample_data$sample_type == "Sputum" | sample_data$sample_type == "BAL" | sample_data$sample_type == "Nasal")


med.fit <- lmer(log10(Final_reads) ~ lypma +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$lypma == 1) %>% subset(., !is.na(.$S.obs)))

out.fit <- lmer(S.obs ~ lypma * log10(Final_reads) + sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$lypma == 1) %>% subset(., !is.na(.$S.obs)))
set.seed(seed)

med.out.lypma <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "lypma",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.lypma)

final.out.lypma <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## benzonase


med.fit <- lmer(log10(Final_reads) ~ benzonase + sample_type + (1|subject_id),
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$benzonase == 1) %>% subset(., !is.na(.$S.obs)))

out.fit <- lmer(S.obs ~ benzonase * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$benzonase == 1) %>% subset(., !is.na(.$S.obs)))

set.seed(seed)
med.out.benzonase <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "benzonase",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.benzonase)

final.out.benzonase <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## host zero


med.fit <- lmer(log10(Final_reads) ~ host_zero +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$host_zero == 1) %>% subset(., !is.na(.$S.obs)))

out.fit <- lmer(S.obs ~ host_zero * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$host_zero == 1) %>% subset(., !is.na(.$S.obs)))
set.seed(seed)
med.out.hostzero <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "host_zero",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.hostzero)

final.out.host_zero <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))


## molysis


med.fit <- lmer(log10(Final_reads) ~ molysis +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$molysis == 1) %>% subset(., !is.na(.$S.obs)))

out.fit <- lmer(S.obs ~ molysis * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$molysis == 1) %>% subset(., !is.na(.$S.obs)))
set.seed(seed)
med.out.molysis <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "molysis",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.molysis)

final.out.molysis <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## qiaamp


med.fit <- lmer(log10(Final_reads) ~ qiaamp +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$qiaamp == 1) %>% subset(., !is.na(.$S.obs)))

out.fit <- lmer(S.obs ~ qiaamp * log10(Final_reads) + sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory_path, sample_data_respiratory_path$control == 1 | sample_data_respiratory_path$qiaamp == 1) %>% subset(., !is.na(.$S.obs)))

set.seed(seed)
med.out_qiaamp <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "qiaamp",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out_qiaamp)

final.out.qiaamp <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

Raw mediation outputs - lyPMA

med.out.lypma %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value   
## ACME (control)             3.6749      -9.1683        21.04  0.5796   
## ACME (treated)            10.6970     -24.4128        49.74  0.5432   
## ADE (control)             46.5096       9.7044        84.73  0.0160 * 
## ADE (treated)             53.5317      15.7030        95.60  0.0072 **
## Total Effect              57.2066      13.7040       103.12  0.0116 * 
## Prop. Mediated (control)   0.0444      -0.3516         0.34  0.5720   
## Prop. Mediated (treated)   0.1774      -0.9425         0.77  0.5352   
## ACME (average)             7.1859     -16.4883        32.98  0.5426   
## ADE (average)             50.0207      14.5112        87.35  0.0074 **
## Prop. Mediated (average)   0.1109      -0.6557         0.51  0.5346   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Raw mediation outputs - Benzonase

med.out.benzonase %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)            45.2255      17.0243        82.33  0.0004 ***
## ACME (treated)            29.9684      -7.0963        74.74  0.1222    
## ADE (control)             46.3120       1.0350        91.62  0.0456 *  
## ADE (treated)             31.0549      -8.4013        69.79  0.1204    
## Total Effect              76.2804      42.7108       109.74  <2e-16 ***
## Prop. Mediated (control)   0.5866       0.2212         1.16  0.0004 ***
## Prop. Mediated (treated)   0.3868      -0.0966         0.98  0.1222    
## ACME (average)            37.5970      10.4486        72.89  0.0062 ** 
## ADE (average)             38.6835       1.1376        76.87  0.0436 *  
## Prop. Mediated (average)   0.4867       0.1372         0.98  0.0062 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 35 
## 
## 
## Simulations: 10000

Raw mediation outputs - host zero

med.out.hostzero %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)             85.341       36.141       143.39  0.0006 ***
## ACME (treated)             28.353      -13.170        74.36  0.1882    
## ADE (control)              94.030       43.662       144.84  0.0004 ***
## ADE (treated)              37.042      -21.360        93.60  0.2090    
## Total Effect              122.383       89.106       154.94  <2e-16 ***
## Prop. Mediated (control)    0.692        0.289         1.20  0.0006 ***
## Prop. Mediated (treated)    0.230       -0.112         0.60  0.1882    
## ACME (average)             56.847       19.573       100.80  0.0032 ** 
## ADE (average)              65.536       19.159       112.00  0.0072 ** 
## Prop. Mediated (average)    0.461        0.158         0.82  0.0032 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 35 
## 
## 
## Simulations: 10000

Raw mediation outputs - MolYsis

med.out.molysis %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                           Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)            77.69495     27.37358       139.74  0.0010 ***
## ACME (treated)            34.40433      1.04967        74.80  0.0432 *  
## ADE (control)             90.56676     45.36088       140.04  <2e-16 ***
## ADE (treated)             47.27614    -14.33392       105.50  0.1248    
## Total Effect             124.97109     84.62700       166.89  <2e-16 ***
## Prop. Mediated (control)   0.61178      0.22487         1.14  0.0010 ***
## Prop. Mediated (treated)   0.27028      0.00817         0.58  0.0432 *  
## ACME (average)            56.04964     23.00149        96.45  <2e-16 ***
## ADE (average)             68.92145     23.70209       113.47  0.0032 ** 
## Prop. Mediated (average)   0.44103      0.19069         0.77  <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 35 
## 
## 
## Simulations: 10000

Raw mediation outputs - qiaamp

med.out_qiaamp %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)             89.667       28.274       159.25   0.003 ** 
## ACME (treated)             77.261       20.186       141.85   0.008 ** 
## ADE (control)              35.405      -27.414        98.75   0.264    
## ADE (treated)              22.999      -46.124        92.26   0.511    
## Total Effect              112.666       75.032       149.76  <2e-16 ***
## Prop. Mediated (control)    0.793        0.246         1.50   0.003 ** 
## Prop. Mediated (treated)    0.686        0.186         1.29   0.008 ** 
## ACME (average)             83.464       32.645       141.73   0.001 ***
## ADE (average)              29.202      -28.441        87.15   0.320    
## Prop. Mediated (average)    0.740        0.284         1.31   0.001 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Tidy table of mediation analysis result (for functional richness)

tableS4_b <- rbind(final.out.lypma,
      final.out.benzonase,
      final.out.host_zero, 
      final.out.molysis,
      final.out.qiaamp) %>%
mutate(Treatment = c("lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"), .before = "Indirect est.") 

tableS4_b

Table S4. Mediation analysis for viruses

outcome = V.obs exposure = treatment (stratified) mediator = Final_reads mediator-outcome confounders = sample_type exposure-mediator confounders = NA outcome model = Mixed effects linear regression mediator model = Mixed effects linear regression

Mediation analysis was conducted stratified treatment.

Table S4. Mediation analysis results in estimated effect sizes and p-values of indirect effect, direct effect, and proportion of mediation. The analysis employed treatment as exposure, log10(final reads) as mediator, species richness as outcome, and sample type as mediator-exposure confounder. Analysis was stratified by each treatment and treated as binary variables.

## lypma
# only mediator-outcome confounders


detach_package(lmerTest) #lmerTest induces error

## all treatment groups

sample_data_respiratory <- subset(phyloseq_unfiltered$phyloseq_count %>% 
        subset_samples(#S.obs != 0 & 
                               sample_type %in% c("Mock", 
                                                       "BAL", "Nasal", "Sputum")) %>% sample_data %>% data.frame, 
                                  sample_data$sample_type == "Sputum" | sample_data$sample_type == "BAL" | sample_data$sample_type == "Nasal") %>%
        subset(., !is.na(.$V.obs))


med.fit <- lmer(log10(Final_reads) ~ lypma +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$lypma == 1))

out.fit <- lmer(V.obs ~ lypma * log10(Final_reads) + sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$lypma == 1))
set.seed(seed)
med.out.lypma_v <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "lypma",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.lypma_v)

final.out.lypma_v <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## benzonase


med.fit <- lmer(log10(Final_reads) ~ benzonase + sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$benzonase == 1))

out.fit <- lmer(V.obs ~ benzonase * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$benzonase == 1))

set.seed(seed)
med.out.benzonase_v <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "benzonase",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.benzonase_v)

final.out.benzonase_v <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## host zero


med.fit <- lmer(log10(Final_reads) ~ host_zero +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$host_zero == 1))

out.fit <- lmer(V.obs ~ host_zero * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$host_zero == 1))
set.seed(seed)
med.out.hostzero_v <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "host_zero",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.hostzero_v)

final.out.host_zero_v <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))


## molysis


med.fit <- lmer(log10(Final_reads) ~ molysis +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$molysis == 1))

out.fit <- lmer(V.obs ~ molysis * log10(Final_reads) +  sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$molysis == 1))
set.seed(seed)
med.out.molysis_v <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "molysis",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out.molysis_v)

final.out.molysis_v <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

## qiaamp


med.fit <- lmer(log10(Final_reads) ~ qiaamp +  sample_type + (1|subject_id),
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$qiaamp == 1))

out.fit <- lmer(V.obs ~ qiaamp * log10(Final_reads) + sample_type + (1|subject_id), 
                data = subset(sample_data_respiratory, sample_data_respiratory$control == 1 | sample_data_respiratory$qiaamp == 1))

set.seed(seed)
med.out_qiaamp_v <- mediate(model.m = med.fit,
                   model.y = out.fit,
                     treat = "qiaamp",
                     mediator = "log10(Final_reads)",
                     sims = 10000)

out.sum <- summary(med.out_qiaamp_v)

final.out.qiaamp_v <- data.frame(`Indirect est.`=out.sum$d.avg, 
                          `Indirect p-value`=out.sum$d.avg.p,
                          `Direct est.`=out.sum$z.avg, 
                          `Direct p-value`=out.sum$z.avg.p, 
                          `Total est.`=out.sum$tau.coef, 
                          `Total p-value`=out.sum$tau.p,
                          `Proportion mediation est.`=out.sum$n.avg, 
                          `Proportion mediation p-value`=out.sum$n.avg.p,
                          check.names = F) %>% 
    mutate_all(~round(., 3))

Raw mediation outputs - lyPMA

med.out.lypma_v %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value
## ACME (control)            0.36552     -2.20995         3.11    0.77
## ACME (treated)            0.43085     -2.66639         3.83    0.77
## ADE (control)             0.00651     -3.60580         3.84    1.00
## ADE (treated)             0.07185     -3.57574         3.97    0.99
## Total Effect              0.43736     -4.10180         5.25    0.87
## Prop. Mediated (control)  0.28714     -5.44982         5.15    0.61
## Prop. Mediated (treated)  0.35765     -5.99530         6.05    0.61
## ACME (average)            0.39819     -2.38467         3.27    0.77
## ADE (average)             0.03918     -3.51113         3.79    0.99
## Prop. Mediated (average)  0.32239     -5.78991         5.48    0.61
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Raw mediation outputs - Benzonase

med.out.benzonase_v %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value   
## ACME (control)             6.2073       1.3430        12.58  0.0084 **
## ACME (treated)             7.0702       0.3184        15.66  0.0392 * 
## ADE (control)              0.1338      -8.9771         8.95  0.9776   
## ADE (treated)              0.9968      -7.0311         9.17  0.8114   
## Total Effect               7.2040      -0.0368        14.46  0.0510 . 
## Prop. Mediated (control)   0.8296      -0.6458         4.48  0.0594 . 
## Prop. Mediated (treated)   0.9435      -0.7582         4.99  0.0862 . 
## ACME (average)             6.6387       1.9651        12.82  0.0042 **
## ADE (average)              0.5653      -7.1233         8.11  0.8794   
## Prop. Mediated (average)   0.8865      -0.5962         4.65  0.0552 . 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Raw mediation outputs - host zero

med.out.hostzero_v %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value  
## ACME (control)             26.759       -6.716        63.97   0.118  
## ACME (treated)             37.689        7.249        72.76   0.014 *
## ADE (control)              -6.555      -43.588        29.55   0.723  
## ADE (treated)               4.375      -36.636        45.37   0.843  
## Total Effect               31.134        6.855        55.42   0.013 *
## Prop. Mediated (control)    0.848       -0.282         3.91   0.129  
## Prop. Mediated (treated)    1.203        0.187         4.69   0.028 *
## ACME (average)             32.224        6.937        61.78   0.012 *
## ADE (average)              -1.090      -34.544        32.22   0.958  
## Prop. Mediated (average)    1.025        0.164         4.04   0.025 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Raw mediation outputs - MolYsis

med.out.molysis_v %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value    
## ACME (control)             18.706       -4.596        46.80  0.1250    
## ACME (treated)             38.091       16.179        65.54  0.0002 ***
## ADE (control)               2.771      -23.259        28.08  0.8294    
## ADE (treated)              22.157       -8.813        56.07  0.1698    
## Total Effect               40.862       17.826        64.84  0.0004 ***
## Prop. Mediated (control)    0.455       -0.119         1.32  0.1254    
## Prop. Mediated (treated)    0.931        0.419         1.96  0.0006 ***
## ACME (average)             28.398       10.946        49.89  0.0002 ***
## ADE (average)              12.464      -11.556        37.28  0.3162    
## Prop. Mediated (average)    0.693        0.280         1.49  0.0006 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Raw mediation outputs - qiaamp

med.out_qiaamp_v %>% summary

## 
## Causal Mediation Analysis 
## 
## Quasi-Bayesian Confidence Intervals
## 
## Mediator Groups: subject_id 
## 
## Outcome Groups: subject_id 
## 
## Output Based on Overall Averages Across Groups 
## 
##                          Estimate 95% CI Lower 95% CI Upper p-value  
## ACME (control)             23.212       -1.092        49.91   0.063 .
## ACME (treated)             28.939        5.100        55.48   0.017 *
## ADE (control)              -8.694      -36.546        18.77   0.532  
## ADE (treated)              -2.967      -31.509        25.26   0.839  
## Total Effect               20.245        4.309        36.14   0.013 *
## Prop. Mediated (control)    1.138       -0.121         4.94   0.076 .
## Prop. Mediated (treated)    1.423        0.178         5.64   0.030 *
## ACME (average)             26.076        6.202        48.54   0.010 *
## ADE (average)              -5.831      -30.260        18.47   0.635  
## Prop. Mediated (average)    1.281        0.232         5.11   0.023 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Sample Size Used: 34 
## 
## 
## Simulations: 10000

Tidy table of mediation analysis result

tableS4_c <- rbind(final.out.lypma_v,
      final.out.benzonase_v,
      final.out.host_zero_v, 
      final.out.molysis_v,
      final.out.qiaamp_v) %>%
mutate(Treatment = c("lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"), .before = "Indirect est.") 

tableS4_c

Mediation analsyis - combined table (Species and functional richness)

tables4 <- rbind (data.frame(Outcome = "Microbial species richness",
           tableS4_a),
       data.frame(Outcome = "Functional richness",
           tableS4_b)
       ) %>% 
        kbl(format = "html") %>%
        kable_styling(full_width = 0, html_font = "sans")

save_kable(tables4, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS4.html", self_contained = T)



tables4_updated <- rbind (data.frame(Outcome = "Microbial species richness",
           tableS4_a),
       data.frame(Outcome = "Functional richness",
           tableS4_b),
       data.frame(Outcome = "Viral species richness",
           tableS4_c)
       ) %>%
        kbl(format = "html") %>%
        kable_styling(full_width = 0, html_font = "sans")

save_kable(tables4_updated, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS4_updated.html", self_contained = T)

Factors affecting DA taxa (main text)

setwd("Project_SICAS2_microbiome/5_Scripts/MGK/Host_depletion_git")

filt_maaslin_all <- read.csv("data/da_lmer_filt_maaslin_all.csv")
#filt_maaslin_interaction <- read.csv("data/da_lmer_filt_maaslin_interaction.csv")
filt_fit_data_bal <- read.csv("data/da_lmer_filt_fit_data_bal.csv")
filt_fit_data_spt <- read.csv("data/da_lmer_filt_fit_data_spt.csv")
filt_fit_data_ns <- read.csv("data/da_lmer_filt_fit_data_ns.csv")
#filt_fit_data_pos <- read.csv("data/da_lmer_filt_fit_data_pos.csv")

MaAsLin output (raw data)

filt_maaslin_all

MaAsLin output (stratified)

BAL stratified MaAsLin result

filt_fit_data_bal

Nasal stratified MaAsLin result

filt_fit_data_ns

Sputum stratified MaAsLin result

filt_fit_data_spt

Number of taxa on some conditions

These calculations were made to write the details in the main text

cat("Factors affecting DA taxa (q<0.1)")

## Factors affecting DA taxa (q<0.1)

filt_maaslin_all %>% subset(., .$qval < 0.1 ) %>% subset(., abs(.$Estimate) > 1) %>% .$metadata %>% table

## .
##   benzonase   host_zero       lypma     molysis      qiaamp sample_type 
##          27          75          25          92          57         145

filt_maaslin_all %>% subset(., .$qval < 0.1 ) %>% .$metadata %>% table

## .
##   benzonase   host_zero       lypma     molysis      qiaamp sample_type 
##          81          90          27          97          92         152

cat("Number of positive & significant changes")

## Number of positive & significant changes

filt_maaslin_all %>% subset(., .$qval < 0.1 ) %>% subset(., .$Estimate > 0) %>% .$feature %>% unique

##  [1] "Actinomyces_graevenitzii"            "Actinomyces_johnsonii"              
##  [3] "Actinomyces_massiliensis"            "Actinomyces_naeslundii"             
##  [5] "Actinomyces_odontolyticus"           "Actinomyces_oris"                   
##  [7] "Actinomyces_sp_HMSC035G02"           "Actinomyces_sp_ICM47"               
##  [9] "Actinomyces_sp_S6_Spd3"              "Actinomyces_sp_oral_taxon_170"      
## [11] "Actinomyces_sp_oral_taxon_180"       "Actinomyces_sp_oral_taxon_181"      
## [13] "Actinomyces_viscosus"                "Aeriscardovia_aeriphila"            
## [15] "Scardovia_wiggsiae"                  "Corynebacterium_accolens"           
## [17] "Corynebacterium_atypicum"            "Corynebacterium_durum"              
## [19] "Corynebacterium_pseudogenitalium"    "Rothia_aeria"                       
## [21] "Rothia_dentocariosa"                 "Rothia_mucilaginosa"                
## [23] "Cutibacterium_acnes"                 "Cutibacterium_granulosum"           
## [25] "Propionibacterium_humerusii"         "Propionibacterium_namnetense"       
## [27] "Pseudopropionibacterium_propionicum" "Atopobium_parvulum"                 
## [29] "Atopobium_rimae"                     "Olsenella_scatoligenes"             
## [31] "Collinsella_aerofaciens"             "Collinsella_intestinalis"           
## [33] "Collinsella_stercoris"               "X.Collinsella._massiliensis"        
## [35] "Slackia_exigua"                      "Slackia_isoflavoniconvertens"       
## [37] "Thermoleophilum_album"               "Porphyromonas_somerae"              
## [39] "Gemella_bergeri"                     "Gemella_haemolysans"                
## [41] "Gemella_morbillorum"                 "Gemella_sanguinis"                  
## [43] "Staphylococcus_argenteus"            "Staphylococcus_epidermidis"         
## [45] "Staphylococcus_schweitzeri"          "Abiotrophia_defectiva"              
## [47] "Dolosigranulum_pigrum"               "Granulicatella_adiacens"            
## [49] "Granulicatella_elegans"              "Streptococcus_anginosus"            
## [51] "Streptococcus_australis"             "Streptococcus_gordonii"             
## [53] "Streptococcus_infantis"              "Streptococcus_mitis"                
## [55] "Streptococcus_oralis"                "Streptococcus_parasanguinis"        
## [57] "Streptococcus_peroris"               "Streptococcus_pneumoniae"           
## [59] "Streptococcus_pseudopneumoniae"      "Streptococcus_salivarius"           
## [61] "Streptococcus_sanguinis"             "Streptococcus_sp_A12"               
## [63] "Streptococcus_sp_F0442"              "Streptococcus_sp_HPH0090"           
## [65] "Streptococcus_sp_M334"               "Streptococcus_vestibularis"         
## [67] "Eubacterium_brachy"                  "Eubacterium_infirmum"               
## [69] "Eubacterium_sulci"                   "Mogibacterium_diversum"             
## [71] "Lachnoanaerobaculum_saburreum"       "Oribacterium_sinus"                 
## [73] "Bulleidia_extructa"                  "Solobacterium_moorei"               
## [75] "Limnochorda_pilosa"                  "Veillonella_dispar"                 
## [77] "Veillonella_parvula"                 "Parvimonas_micra"                   
## [79] "Fusobacterium_nucleatum"             "Cupriavidus_sp"                     
## [81] "Sutterella_parvirubra"               "Pseudomonas_aeruginosa"             
## [83] "Malassezia_restricta"                "Neisseria_flavescens"               
## [85] "Stenotrophomonas_maltophilia"

cat("Number of negative & significant changes")

## Number of negative & significant changes

filt_maaslin_all %>% subset(., .$qval < 0.1 ) %>% subset(., .$Estimate < 0) %>% .$feature %>% unique

##   [1] "Actinomyces_georgiae"                
##   [2] "Actinomyces_hongkongensis"           
##   [3] "Actinomyces_sp_oral_taxon_414"       
##   [4] "Corynebacterium_accolens"            
##   [5] "Corynebacterium_atypicum"            
##   [6] "Corynebacterium_pseudodiphtheriticum"
##   [7] "Corynebacterium_pseudogenitalium"    
##   [8] "Corynebacterium_tuberculostearicum"  
##   [9] "Cutibacterium_acnes"                 
##  [10] "Cutibacterium_granulosum"            
##  [11] "Propionibacterium_humerusii"         
##  [12] "Propionibacterium_namnetense"        
##  [13] "Atopobium_deltae"                    
##  [14] "Atopobium_parvulum"                  
##  [15] "Asaccharobacter_celatus"             
##  [16] "Denitrobacterium_detoxificans"       
##  [17] "Enterorhabdus_caecimuris"            
##  [18] "Rubrobacter_radiotolerans"           
##  [19] "Porphyromonas_endodontalis"          
##  [20] "Prevotella_histicola"                
##  [21] "Prevotella_melaninogenica"           
##  [22] "Prevotella_oris"                     
##  [23] "Kouleothrix_aurantiaca"              
##  [24] "Hydrogenibacillus_schlegelii"        
##  [25] "Gemella_asaccharolytica"             
##  [26] "Brochothrix_campestris"              
##  [27] "Staphylococcus_epidermidis"          
##  [28] "Dolosigranulum_pigrum"               
##  [29] "Enterococcus_faecalis"               
##  [30] "Lactobacillus_rhamnosus"             
##  [31] "Streptococcus_anginosus"             
##  [32] "Streptococcus_parasanguinis"         
##  [33] "Eubacterium_infirmum"                
##  [34] "Oribacterium_parvum"                 
##  [35] "Oribacterium_sp_oral_taxon_078"      
##  [36] "Stomatobaculum_longum"               
##  [37] "Veillonella_atypica"                 
##  [38] "Veillonella_dispar"                  
##  [39] "Veillonella_infantium"               
##  [40] "Veillonella_sp_T11011_6"             
##  [41] "Anaerococcus_nagyae"                 
##  [42] "Finegoldia_magna"                    
##  [43] "Fusobacterium_nucleatum"             
##  [44] "Leptotrichia_wadei"                  
##  [45] "Gemmata_obscuriglobus"               
##  [46] "Paludisphaera_borealis"              
##  [47] "Achromobacter_xylosoxidans"          
##  [48] "Cupriavidus_sp"                      
##  [49] "Sutterella_parvirubra"               
##  [50] "Cardiobacterium_valvarum"            
##  [51] "Alkalilimnicola_ehrlichii"           
##  [52] "Escherichia_coli"                    
##  [53] "Thiohalorhabdus_denitrificans"       
##  [54] "Pseudomonas_aeruginosa"              
##  [55] "Acholeplasma_oculi"                  
##  [56] "Candida_albicans"                    
##  [57] "Candida_dubliniensis"                
##  [58] "Malassezia_restricta"                
##  [59] "Nakamurella_silvestris"              
##  [60] "Alloprevotella_tannerae"             
##  [61] "Prevotella_jejuni"                   
##  [62] "Prevotella_pallens"                  
##  [63] "Prevotella_sp_oral_taxon_306"        
##  [64] "Prevotella_veroralis"                
##  [65] "Tannerella_sp_oral_taxon_808"        
##  [66] "Bacillus_ginsengihumi"               
##  [67] "Listeria_floridensis"                
##  [68] "Agitococcus_lubricus"                
##  [69] "Lactobacillus_fermentum"             
##  [70] "Lactobacillus_gasseri"               
##  [71] "Sharpea_azabuensis"                  
##  [72] "Oribacterium_asaccharolyticum"       
##  [73] "Stenotrophomonas_maltophilia"        
##  [74] "Stenotrophomonas_pavanii"            
##  [75] "Stenotrophomonas_rhizophila"         
##  [76] "Bacillus_intestinalis"               
##  [77] "Listeria_innocua"                    
##  [78] "Listeria_monocytogenes"              
##  [79] "Staphylococcus_haemolyticus"         
##  [80] "Eggerthia_catenaformis"              
##  [81] "Leptotrichia_sp_oral_taxon_498"      
##  [82] "Rickettsia_felis"                    
##  [83] "Haemophilus_parainfluenzae"          
##  [84] "Candida_orthopsilosis"               
##  [85] "Nannocystis_exedens"                 
##  [86] "Campylobacter_concisus"              
##  [87] "Franconibacter_helveticus"           
##  [88] "Klebsiella_pneumoniae"               
##  [89] "Salmonella_enterica"                 
##  [90] "Superficieibacter_electus"           
##  [91] "Pseudomonas_fluorescens"             
##  [92] "Saccharomyces_cerevisiae"            
##  [93] "S._cerevisiae.x.S._kudriavzevii"     
##  [94] "Saccharomyces_kudriavzevii"          
##  [95] "Cryptococcus_gattii_VGI"             
##  [96] "Cryptococcus_gattii_VGII"            
##  [97] "Cryptococcus_gattii_VGIII"           
##  [98] "Cryptococcus_neoformans"             
##  [99] "Listeria_marthii"                    
## [100] "Erwinia_persicina"                   
## [101] "Pseudomonas_formosensis"             
## [102] "Pseudomonas_putida"

cat("BAL highly changed taxa (q<0.1 & abs(.$coef) >1)")

## BAL highly changed taxa (q<0.1 & abs(.$coef) >1)

filt_fit_data_bal %>% subset(., .$qval < 0.1 ) %>% subset(., abs(.$Estimate) >1 ) %>% .$metadata %>% table

## < table of extent 0 >

cat("Sputum highly changed taxa (q<0.1 & abs(.$coef) >1)")

## Sputum highly changed taxa (q<0.1 & abs(.$coef) >1)

filt_fit_data_spt %>% subset(., .$qval < 0.1 ) %>% subset(., abs(.$Estimate) >1 ) %>% .$metadata %>% table

## .
## benzonase host_zero     lypma   molysis    qiaamp 
##        86       101        81       102       111

cat("NS highly changed taxa (q<0.1 & abs(.$coef) >1)")

## NS highly changed taxa (q<0.1 & abs(.$coef) >1)

filt_fit_data_ns %>% subset(., .$qval < 0.1 ) %>% subset(., abs(.$Estimate) >1 ) %>% .$metadata %>% table

## .
## benzonase host_zero     lypma   molysis    qiaamp 
##         1         2         7         2         5

cat("NS taxa (q<0.1)")

## NS taxa (q<0.1)

filt_fit_data_bal %>% subset(., .$qval < 0.1 ) %>% .$feature %>% unique

## character(0)

cat("NS taxa (q<0.1)")

## NS taxa (q<0.1)

filt_fit_data_ns %>% subset(., .$qval < 0.1 ) %>% .$feature %>% unique

##  [1] "Actinomyces_odontolyticus"           
##  [2] "Actinomyces_sp_ICM47"                
##  [3] "Aeriscardovia_aeriphila"             
##  [4] "Corynebacterium_accolens"            
##  [5] "Corynebacterium_pseudodiphtheriticum"
##  [6] "Olsenella_scatoligenes"              
##  [7] "X.Collinsella._massiliensis"         
##  [8] "Rubrobacter_radiotolerans"           
##  [9] "Thermoleophilum_album"               
## [10] "Gemella_asaccharolytica"             
## [11] "Staphylococcus_aureus"               
## [12] "Staphylococcus_epidermidis"          
## [13] "Streptococcus_oralis"                
## [14] "Limnochorda_pilosa"                  
## [15] "Finegoldia_magna"                    
## [16] "Cupriavidus_sp"                      
## [17] "Sutterella_parvirubra"               
## [18] "Acholeplasma_oculi"                  
## [19] "Malassezia_restricta"

cat("Sputum taxa (q<0.1 & abs(.$coef) >1)")

## Sputum taxa (q<0.1 & abs(.$coef) >1)

filt_fit_data_spt %>% subset(., .$qval < 0.1 ) %>% .$feature %>% unique

##   [1] "Actinomyces_georgiae"                
##   [2] "Actinomyces_hongkongensis"           
##   [3] "Actinomyces_johnsonii"               
##   [4] "Actinomyces_massiliensis"            
##   [5] "Actinomyces_naeslundii"              
##   [6] "Actinomyces_odontolyticus"           
##   [7] "Actinomyces_oris"                    
##   [8] "Actinomyces_sp_HMSC035G02"           
##   [9] "Actinomyces_sp_HPA0247"              
##  [10] "Actinomyces_sp_ICM47"                
##  [11] "Actinomyces_sp_S6_Spd3"              
##  [12] "Actinomyces_sp_oral_taxon_170"       
##  [13] "Actinomyces_sp_oral_taxon_180"       
##  [14] "Actinomyces_sp_oral_taxon_181"       
##  [15] "Actinomyces_sp_oral_taxon_414"       
##  [16] "Aeriscardovia_aeriphila"             
##  [17] "Alloscardovia_omnicolens"            
##  [18] "Bifidobacterium_dentium"             
##  [19] "Scardovia_wiggsiae"                  
##  [20] "Corynebacterium_accolens"            
##  [21] "Corynebacterium_atypicum"            
##  [22] "Corynebacterium_durum"               
##  [23] "Corynebacterium_matruchotii"         
##  [24] "Corynebacterium_pseudodiphtheriticum"
##  [25] "Rothia_aeria"                        
##  [26] "Rothia_dentocariosa"                 
##  [27] "Cutibacterium_acnes"                 
##  [28] "Propionibacterium_acidifaciens"      
##  [29] "Pseudopropionibacterium_propionicum" 
##  [30] "Atopobium_deltae"                    
##  [31] "Atopobium_parvulum"                  
##  [32] "Atopobium_rimae"                     
##  [33] "Olsenella_scatoligenes"              
##  [34] "Olsenella_uli"                       
##  [35] "Collinsella_aerofaciens"             
##  [36] "Collinsella_stercoris"               
##  [37] "Enorma_massiliensis"                 
##  [38] "X.Collinsella._massiliensis"         
##  [39] "Asaccharobacter_celatus"             
##  [40] "Denitrobacterium_detoxificans"       
##  [41] "Enterorhabdus_caecimuris"            
##  [42] "Slackia_exigua"                      
##  [43] "Slackia_isoflavoniconvertens"        
##  [44] "Rubrobacter_radiotolerans"           
##  [45] "Porphyromonas_endodontalis"          
##  [46] "Porphyromonas_somerae"               
##  [47] "Prevotella_histicola"                
##  [48] "Prevotella_melaninogenica"           
##  [49] "Prevotella_oris"                     
##  [50] "Kouleothrix_aurantiaca"              
##  [51] "Hydrogenibacillus_schlegelii"        
##  [52] "Gemella_asaccharolytica"             
##  [53] "Gemella_bergeri"                     
##  [54] "Gemella_haemolysans"                 
##  [55] "Gemella_morbillorum"                 
##  [56] "Brochothrix_campestris"              
##  [57] "Staphylococcus_argenteus"            
##  [58] "Staphylococcus_epidermidis"          
##  [59] "Staphylococcus_schweitzeri"          
##  [60] "Abiotrophia_defectiva"               
##  [61] "Abiotrophia_sp_HMSC24B09"            
##  [62] "Dolosigranulum_pigrum"               
##  [63] "Granulicatella_adiacens"             
##  [64] "Granulicatella_elegans"              
##  [65] "Enterococcus_faecalis"               
##  [66] "Lactobacillus_rhamnosus"             
##  [67] "Streptococcus_anginosus"             
##  [68] "Streptococcus_australis"             
##  [69] "Streptococcus_gordonii"              
##  [70] "Streptococcus_infantis"              
##  [71] "Streptococcus_mitis"                 
##  [72] "Streptococcus_oralis"                
##  [73] "Streptococcus_parasanguinis"         
##  [74] "Streptococcus_peroris"               
##  [75] "Streptococcus_pneumoniae"            
##  [76] "Streptococcus_pseudopneumoniae"      
##  [77] "Streptococcus_sanguinis"             
##  [78] "Streptococcus_sp_A12"                
##  [79] "Streptococcus_sp_F0442"              
##  [80] "Streptococcus_sp_HMSC034E03"         
##  [81] "Streptococcus_sp_HMSC067H01"         
##  [82] "Streptococcus_sp_HPH0090"            
##  [83] "Streptococcus_sp_M334"               
##  [84] "Eubacterium_brachy"                  
##  [85] "Eubacterium_infirmum"                
##  [86] "Eubacterium_sulci"                   
##  [87] "Mogibacterium_diversum"              
##  [88] "Mogibacterium_pumilum"               
##  [89] "Mogibacterium_timidum"               
##  [90] "Lachnoanaerobaculum_saburreum"       
##  [91] "Oribacterium_parvum"                 
##  [92] "Oribacterium_sinus"                  
##  [93] "Oribacterium_sp_oral_taxon_078"      
##  [94] "Stomatobaculum_longum"               
##  [95] "Peptostreptococcus_stomatis"         
##  [96] "Bulleidia_extructa"                  
##  [97] "Solobacterium_moorei"                
##  [98] "Limnochorda_pilosa"                  
##  [99] "Veillonella_atypica"                 
## [100] "Veillonella_dispar"                  
## [101] "Veillonella_infantium"               
## [102] "Veillonella_sp_T11011_6"             
## [103] "Parvimonas_micra"                    
## [104] "Leptotrichia_wadei"                  
## [105] "Gemmata_obscuriglobus"               
## [106] "Paludisphaera_borealis"              
## [107] "Achromobacter_xylosoxidans"          
## [108] "Cupriavidus_sp"                      
## [109] "Sutterella_parvirubra"               
## [110] "Anaerobiospirillum_thomasii"         
## [111] "Cardiobacterium_hominis"             
## [112] "Cardiobacterium_valvarum"            
## [113] "Alkalilimnicola_ehrlichii"           
## [114] "Escherichia_coli"                    
## [115] "Thiohalorhabdus_denitrificans"       
## [116] "Pseudomonas_aeruginosa"              
## [117] "Acholeplasma_oculi"                  
## [118] "Candida_albicans"                    
## [119] "Candida_dubliniensis"                
## [120] "Candida_parapsilosis"                
## [121] "Nakamurella_silvestris"              
## [122] "Prevotella_jejuni"                   
## [123] "Prevotella_pallens"                  
## [124] "Prevotella_salivae"                  
## [125] "Prevotella_sp_oral_taxon_306"        
## [126] "Prevotella_veroralis"                
## [127] "Tannerella_sp_oral_taxon_HOT_286"    
## [128] "Capnocytophaga_gingivalis"           
## [129] "Listeria_floridensis"                
## [130] "Agitococcus_lubricus"                
## [131] "Lactobacillus_fermentum"             
## [132] "Lactobacillus_gasseri"               
## [133] "Sharpea_azabuensis"                  
## [134] "Oribacterium_asaccharolyticum"       
## [135] "Megasphaera_micronuciformis"         
## [136] "Neisseria_flavescens"                
## [137] "Neisseria_subflava"                  
## [138] "Stenotrophomonas_maltophilia"        
## [139] "Stenotrophomonas_pavanii"            
## [140] "Stenotrophomonas_rhizophila"         
## [141] "Eggerthia_catenaformis"              
## [142] "Haemophilus_parainfluenzae"          
## [143] "Candida_orthopsilosis"               
## [144] "Nannocystis_exedens"                 
## [145] "Campylobacter_concisus"

cat("Sputum taxa (q<0.1 & abs(.$coef) >1)")

## Sputum taxa (q<0.1 & abs(.$coef) >1)

filt_fit_data_spt %>% subset(., .$qval < 0.1 ) %>% .$metadata %>% table

## .
## benzonase host_zero     lypma   molysis    qiaamp 
##        86       101        82       102       111

Fig. S8. Volcano plot

Fig. S8. Volcano plot of differential abundance of microbes by each treatment with a model MaAsLin (relative abundance of each taxa ~ sample type + lyPMA + Benzonase + HostZERO + MolYsis + QIAamp, random effect = subject id).

#Making significance table for figure
        # Define a function to make species names italicized
# Make a significance table for each figure (top 20 taxa)
species_italic <- function(data) {
  names <- gsub("_", " ", rownames(data))
  names <- gsub("[]]|[[]", "", names)
  names <- gsub(" sp", " sp.", names)
  names <- gsub(" sp.", "* sp.", names)
  names <- gsub(" group", "", names)
  names <- ifelse(grepl("[*]", names), paste("*", names, sep = ""), paste("*", names, "*", sep = ""))
  rownames(data) <- names
  data
}
species_revise <- function(data) {
  data$feature <- gsub("Saccharomyces_cerevisiae_x_Saccharomyces_kudriavzevii", "S._cerevisiae x S._kudriavzevii", data$feature)
  data$feature <- gsub("Pseudomonas_aeruginosa_group", "Pseudomonas_aeruginosa", data$feature)
  data$feature <- gsub("Pseudomonas_fluorescens_group", "Pseudomonas_fluorescens", data$feature)
  data
}



make_sig_table <- function(data) {
  sig_data <- spread(data[order(data$qval), c("feature", "metadata", "qval")], metadata, qval)
  sig_data <- species_revise(sig_data)
  sig_data$min <- apply(sig_data %>% dplyr::select(c("lypma", "benzonase", "molysis", "host_zero", "qiaamp")), 1, FUN = min)
  sig_data <- sig_data[order(sig_data$min),] %>% dplyr::select("feature", "lypma", "benzonase", "host_zero", "molysis", "qiaamp") %>% .[1:20,]
  sig_data[["feature"]] <- ifelse(sig_data[["feature"]] == "X.Collinsella._massiliensis", "[Collinsella]_massiliensis", sig_data[["feature"]])
  sig_data_italic <- sig_data %>% rownames_to_column(var = "-") %>%
          column_to_rownames(var = "feature") %>% species_italic %>% dplyr::select(-c("-")) %>%
          rename(lyPMA = lypma,  Benzonase = benzonase, `HostZERO` = host_zero, MolYsis = molysis, QIAamp = qiaamp)
  sig_data_sig <- ifelse(sig_data_italic < 0.1, "*", NA) %>% data.frame(check.names = F)
  return(list(data = sig_data, data_italic = sig_data_italic, data_sig = sig_data_sig))
}



#filt_fit_data_pos <- make_sig_table(filt_fit_data_pos)
filt_fit_data_bal <- make_sig_table(filt_fit_data_bal)
filt_fit_data_ns <- make_sig_table(filt_fit_data_ns)
filt_fit_data_spt <- make_sig_table(filt_fit_data_spt)

#filt_pos_sig <- subset_taxa(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "Mock"),
#                                       taxa_names(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "Mock")) %in% filt_fit_data_pos$data$feature)

#filt_fit_data_pos$rel <- cbind(filt_pos_sig %>% otu_table %>% t, filt_pos_sig %>% sample_data) %>% group_by(treatment) %>% summarise_if(is.numeric, mean, na.rm = TRUE) %>% .[, 1:21] %>% column_to_rownames(., "treatment") %>% t () %>% species_italic() %>%  data.frame(check.names = F) %>% 
#        .[row.names(filt_fit_data_pos$data_italic),] %>%  mutate_all(~na_if(., 0)) %>% rownames_to_column("feature")

filt_spt_sig <- subset_taxa(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "Sputum"), taxa_names(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "Sputum")) %in% filt_fit_data_spt$data$feature)

filt_fit_data_spt$rel <- cbind(filt_spt_sig %>% otu_table %>% t, filt_spt_sig %>% sample_data) %>% group_by(treatment) %>% summarise_if(is.numeric, mean, na.rm = TRUE) %>% .[, 1:21] %>% column_to_rownames(., "treatment") %>% t () %>% species_italic() %>%  data.frame(check.names = F) %>% 
        .[row.names(filt_fit_data_spt$data_italic),] %>%  mutate_all(~na_if(., 0)) %>% rownames_to_column("feature")

filt_ns_sig <- subset_taxa(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "Nasal"),
                                       taxa_names(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "Nasal")) %in% filt_fit_data_ns$data$feature)

filt_fit_data_ns$rel <- cbind(filt_ns_sig %>% otu_table %>% t, filt_ns_sig %>% sample_data) %>% group_by(treatment) %>% summarise_if(is.numeric, mean, na.rm = TRUE) %>% .[, 1:21] %>% column_to_rownames(., "treatment") %>% t () %>% species_italic() %>%  data.frame(check.names = F) %>% 
        .[row.names(filt_fit_data_ns$data_italic),] %>%  mutate_all(~na_if(., 0)) %>% rownames_to_column("feature")

filt_bal_sig <- subset_taxa(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "BAL"),
                                       taxa_names(subset_samples(phyloseq_unfiltered$phyloseq_rel, sample_type == "BAL")) %in% filt_fit_data_bal$data$feature)

filt_fit_data_bal$rel <- cbind(filt_bal_sig %>% otu_table %>% t, filt_bal_sig %>% sample_data) %>% group_by(treatment) %>% summarise_if(is.numeric, mean, na.rm = TRUE) %>% .[, 1:21] %>% column_to_rownames(., "treatment") %>% t () %>% species_italic() %>%  data.frame(check.names = F) %>%
        .[row.names(filt_fit_data_bal$data_italic),] %>%
        mutate_all(~na_if(., 0)) %>% rownames_to_column("feature") %>% subset(., !grepl("NA", .$feature))

#Volcano plot

figS7 <- ggplot(filt_maaslin_all, aes(y = -log10(qval), x = Estimate, col = metadata)) +
        theme_classic(base_family = "sans") +
        #labs(tag = "A") +
        geom_point(size = 2, alpha = 0.3, stroke = 0) +
        xlab("Change estimate of CLR-normalized read counts") +
        ylab("-log<sub>10</sub>(*q*-value)") +
        #ylim(c(-1, 35)) +
        geom_hline(yintercept = 1, col = "gray") +
        geom_vline(xintercept = 0, col = "gray") +
        annotate(family = "sans",
                 geom='richtext',
                 x=0, y=20,
                 label = "<i>q</i>-value = 0.1, fold-change = 0") +
        theme(legend.position = "top", axis.title.y = ggtext::element_markdown(), legend.text = element_markdown()) +
        scale_color_manual(values = c("#a65628",
                                      "grey",
                                      "#fb9a99",
                                      "#33a02c",
                                      "#b2df8a",
                                      "#1f78b4",
                                      "#a6cee3"),
                           breaks = c("log10.Final_reads",
                                      "sample_type",
                                      "lypma",
                                      "benzonase",
                                      "host_zero",
                                      "molysis",
                                      "qiaamp"), 
                           labels = c("log<sub>10</sub>(Final reads)",
                                      "Sample type",
                                      "lyPMA",
                                      "Benzonase",
                                      "HostZERO",
                                      "MolYsis",
                                      "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        guides(col = guide_legend(title = "Factors", title.position = "top", nrow = 1))

figS7

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS8.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 90, # The height of the plot in inches/
    units = "mm",
    res = 600
) #fixing multiple page issue
figS7
dev.off()

## quartz_off_screen 
##                 2

Fig. 4. Balloon plot

Fig. 4. Mean relative abundance of top 20 significant taxa by q-value identified by differential abundance analysis using MaAsLin. Analyses were stratified by sample type. (A) bronchoalveolor lavage, (B) nasal swabs, and (C) sputum. Statistical significances were noted at the level of q-value < 0.1

#ffff33 qia

f5b <- merge(filt_fit_data_bal$rel %>%
              gather(treatment,
                     value,
                     Untreated:QIAamp,
                     factor_key=TRUE),
      filt_fit_data_bal$data_italic %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     qval,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
      by.x = c('feature', 'treatment'),
      by.y = c('feature', 'treatment'),
      all = T) %>%
        
        merge(filt_fit_data_bal$data_sig %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     sig,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
              by.x = c('feature', 'treatment'),
              by.y = c('feature', 'treatment'),
              all = T) %>%
        mutate(sig = case_when(is.na(qval) ~ "> 0.1",
                               qval < 0.1 ~ "< 0.1",
                               .default = "> 0.1")) %>%

#Baloon plot
        ggballoonplot(size = "value", y = "feature", x= "treatment", fill = "sig") +
        theme_classic(base_family = "sans") +
        #colors for qvalues
        #xlab("Experimental group") +
        #ylab("Species") +
        labs(tag = "A")  +
        ggtitle("BAL") +
        theme(panel.grid.major = element_line(colour = "grey"),
              legend.position = "top",
              #axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
              #Element markdown for taxa name italicizing
              axis.text.y = ggtext::element_markdown(size = 8),
              axis.title.y = element_blank(),
              axis.title.x = element_blank(),
              plot.margin = unit(c(0,0.2,0,1), 'lines'))  +
        scale_fill_manual(values = c("grey", "red"), aes(y = feature,
                      x = treatment,
                      label = sig)) +
        guides(fill = guide_legend(title = c(expression(paste(italic("q"),
                                                       "-value",
                                                       sep = ""))),
                                   title.position = "top"),
               size = guide_legend(title = "Relative abundance",
                                   title.position = "top",
                                   order = 1,
                                   nrow = 1)
               ) + 
        scale_x_discrete(labels=c("control" = "Untreated",
                                  "lypma" = "lyPMA",
                                  "benzonase" = "Benzonase",
                                  "host_zero" = "Host-zero",
                                  "molysis" = "MolYsis",
                                  "qiaamp" = "QIAamp")
                         )

f5c <- merge(filt_fit_data_ns$rel %>%
              gather(treatment,
                     value,
                     Untreated:QIAamp,
                     factor_key=TRUE),
      filt_fit_data_ns$data_italic %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     qval,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
      by.x = c('feature', 'treatment'),
      by.y = c('feature', 'treatment'),
      all = T) %>%
        
        merge(filt_fit_data_ns$data_sig %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     sig,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
              by.x = c('feature', 'treatment'),
              by.y = c('feature', 'treatment'),
              all = T) %>%
        mutate(sig = case_when(sig < 0.1 ~ "< 0.1",
                               .default = "> 0.1")) %>%
#Baloon plot
        ggballoonplot(size = "value", y = "feature", x= "treatment", fill = "sig") +
        
        theme_classic(base_family = "sans") +
        #colors for qvalues
        gradient_fill(c("#006d2c", "#edf8fb")) +
        xlab("Experimental group") +
        ylab("Species") +
        labs(tag = "B")  +
        ggtitle("Nasal swab") +
        theme(panel.grid.major = element_line(colour = "grey"),
              legend.position = "top",
              #axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
              #Element markdown for taxa name italicizing
              axis.text.y = ggtext::element_markdown(size = 8),
              axis.title.x = element_blank(),
              plot.margin = unit(c(0,0.2,0,1), 'lines'))  +
        #Adding significance asterisks
        scale_fill_manual(values = c("red", "grey"), aes(y = feature,
                      x = treatment,
                      label = sig)) +
        guides(fill = guide_legend(title = c(expression(paste(italic("q"),
                                                       "-value",
                                                       sep = ""))),
                                   title.position = "top",
                                   override.aes = list(size=5)),
               size = guide_legend(title = "Relative abundance",
                                   title.position = "top",
                                   order = 1,
                                   nrow = 1),
               ) + 
        scale_x_discrete(labels=c("control" = "Untreated",
                                  "lypma" = "lyPMA",
                                  "benzonase" = "Benzonase",
                                  "host_zero" = "Host-zero",
                                  "molysis" = "MolYsis",
                                  "qiaamp" = "QIAamp")
                         )

f5d <- merge(filt_fit_data_spt$rel %>%
              gather(treatment,
                     value,
                     Untreated:QIAamp,
                     factor_key=TRUE),
      filt_fit_data_spt$data_italic %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     qval,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
      by.x = c('feature', 'treatment'),
      by.y = c('feature', 'treatment'),
      all = T) %>%
        merge(filt_fit_data_spt$data_sig %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     sig,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
              by.x = c('feature', 'treatment'),
              by.y = c('feature', 'treatment'),
              all = T) %>%
        mutate(sig = case_when(sig < 0.1 ~ "< 0.1",
                               .default = "> 0.1")) %>%
#Baloon plot
        ggballoonplot(size = "value", y = "feature", x= "treatment", fill = "sig") +
        
        theme_classic(base_family = "sans") +
        #colors for qvalues
        gradient_fill(c("#006d2c", "#edf8fb")) +
        xlab("Experimental group") +
        #ylab("Species") +
        labs(tag = "C")  +
        ggtitle("Sputum") +
        theme(panel.grid.major = element_line(colour = "grey"),
              legend.position = "top",
              #axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
              #Element markdown for taxa name italicizing
              axis.text.y = ggtext::element_markdown(size = 8),
              #axis.title.x = element_blank(),
              axis.title.x = element_text(margin = margin(t = 0)),
              axis.title.y = element_blank(),
              plot.margin = unit(c(0,0.2,0,1), 'lines'))  +
        #Adding significance asterisks
        scale_fill_manual(values = c("red", "grey"), aes(y = feature,
                      x = treatment,
                      label = sig)) +
        guides(fill = guide_legend(title = c(expression(paste(italic("q"),
                                                       "-value",
                                                       sep = ""))),
                                   title.position = "top"),
               size = guide_legend(title = "Relative abundance",
                                   title.position = "top",
                                   order = 1,
                                   nrow = 1),
               ) + 
        scale_x_discrete(labels=c("control" = "Untreated",
                                  "lypma" = "lyPMA",
                                  "benzonase" = "Benzonase",
                                  "host_zero" = "Host-zero",
                                  "molysis" = "MolYsis",
                                  "qiaamp" = "QIAamp")
                         )

fig4 <- ggarrange(f5c %>% lemon::g_legend() %>% as_ggplot,
                  f5b,
                  f5c,
                  f5d,
                  #b, b %>% lemon::g_legend() %>% as_ggplot,
                  ncol=1, heights = c(1.5, 4, 4, 4),
                  legend = "none",
                  align = "hv")

fig4

# png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/Figure4.png",   # The directory you want to save the file in
#     width = 180, # The width of the plot in inches
#     height = 220, # The height of the plot in inches
#     units = "mm",
#     res = 600
# ) #fixing multiple page issue
# 
# fig4
# 
# 
# 
# 
# # alpha diversity plots
# #ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
# #          ncol = 1) # alpha diversity plots
# 
# dev.off()
# 



pdf(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/Figure4.pdf",   # The directory you want to save the file in
    width = 183 / 25.4, # Convert width from mm to inches (180 mm)
    height = 220 / 25.4, # Convert height from mm to inches (160 mm)
    paper = "special",   # Prevents default paper size settings
    onefile = FALSE      # Ensures a single page per PDF file
)


fig4
# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Table S8. List of DA taxa

Table S8. List of differentially abundant taxa identified by MaAsLin (Beghini et al., 2021). Significant associations (q-value < 0.1) that is showing high change ( | log2(fold-change) | > 0.5) were listed.

filt_maaslin_all %>% subset(., .$qval < 0.1 & abs(.$Estimate) > 4) 

tableS8 <- filt_maaslin_all %>% subset(., .$qval < 0.1 & abs(.$Estimate) > 4) %>% dplyr::select(c("feature", "metadata", "Estimate", "qval")) %>%
        mutate(feature = paste("<i>", gsub("_", " ", feature), sep = "") %>%
                       gsub(" sp", "</i> sp.", .) %>% gsub(" group", "", .)) %>%
        mutate(feature = case_when(!grepl("</i>", feature) ~ paste(feature, "</i>", sep = ""),
                                   .default = feature))%>% 
        rename(Taxa = "feature",
                     `Fixed effect` = "metadata",
                     "Change estimate of CLR transformed count" = "Estimate",
                     `<i>q<i/>-value` = "qval") %>% 
        mutate(`Fixed effect` = case_when(
                         `Fixed effect` == "sample_type" ~ "Sample type",
                         `Fixed effect` == "lypma" ~ "lyPMA",
                         `Fixed effect` == "host_zero" ~ "HostZERO",
                         `Fixed effect` == "benzonase" ~ "Benzonase",
                         `Fixed effect` == "molysis" ~ "MolYsis",
                         `Fixed effect` == "qiaamp" ~ "QIAamp",
                         .default = `Fixed effect`),
               "Change estimate of CLR transformed count" = round(`Change estimate of CLR transformed count`, 3),
               `<i>q<i/>-value` = round(`<i>q<i/>-value`, 3)) %>%
        remove_rownames() %>% 
        kbl(format = "html", escape = FALSE) %>%
        kable_styling(full_width = 0, html_font = "sans")

tableS8

save_kable(tableS8, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS8.html", self_contained = T)

Fig. S9. Figure of alpha and beta

Fig. S9. Alpha and beta diversity by sample type and treatment method of predicted functions. (A) Species richness with statistical test results (linear mixed effect model stratified by sample type), (B) Morisita-Horn dissimilarity within subject between treatment, representing squares for median value and bars for 95% confidence intervals.

sample_data <- sample_data(phyloseq$phyloseq_path_rpk) %>% data.frame(check.names = F) %>% subset(., !is.nan(.$simpson))
phyloseq_rel_nz_f <- subset_samples(phyloseq$phyloseq_path_rpk, S.obs != 0 & sample_type %in% c("BAL", "Nasal", "Sputum", "Mock"))

sample_data(phyloseq_rel_nz_f)$log10.Final_reads <- log10(sample_data(phyloseq_rel_nz_f)$Final_reads)
sample_data(phyloseq_rel_nz_f)$sampletype_treatment <- paste(sample_data(phyloseq_rel_nz_f)$sample_type, sample_data(phyloseq_rel_nz_f)$treatment, sep = ":")

f4a <- ggplot(subset(sample_data(phyloseq$phyloseq_path_rpk) %>% 
                             data.frame,
                     sample_data(phyloseq$phyloseq_path_rpk)$sample_type %in% 
                             c("Sputum","Nasal", "BAL"#, "Mock"
                               )), aes(x = treatment, y = S.obs)) +
        geom_jitter(aes(color = treatment), position = position_jitter(0.2), size = 1.2,
                    alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        ylab("Functional richness") +
        xlab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using 
        labs(tag = "A") +
        theme(plot.tag = element_text(size = 15),  axis.text.x = element_blank(), axis.ticks.x = element_blank()) +
        facet_wrap(~sample_type, nrow = 1) + 
        guides(col = guide_legend(nrow = 1)) 

dat_text <- data.frame(
  label = c(
          #"**", "***", "***", "", "***", #label for Mock
          "", "*", "**", "***", "*", #label for BAL
          "", "", "***", "**", "***", 
          "*", "**", "***", "***", "***"),
  sample_type = c(
          #"Mock", "Mock", "Mock", "Mock", "Mock", 
          "BAL", "BAL", "BAL", "BAL", "BAL", 
          "Nasal", "Nasal", "Nasal", "Nasal", "Nasal", 
          "Sputum", "Sputum", "Sputum", "Sputum", "Sputum"),
  treatment     = c(
          #"lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp", 
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp",
          "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
  S.obs = c(
          #420, 400, 330, 410, 400,
          250, 230, 300, 320, 300,
          210, 220, 230, 240, 250,
          330, 350, 370, 350, 360)
)

dat_text$sample_type <- factor(dat_text$sample_type, levels = c(#"Mock", 
                                                                "BAL", "Nasal", "Sputum"))
dat_text$treatment <- factor(dat_text$treatment, levels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))



f4a <- f4a + geom_text(
  data    = dat_text,
  mapping = aes(x = treatment, y = S.obs, label = label)
)


#f3a <- f3a + geom_text(
#  data    = dat_text,
#  mapping = aes(x = treatment, y = S.obs, label = label)
#)


#distances of betadiversity - boxplots
horn_dist_long <- distance(phyloseq_rel_nz_f, method="horn") %>% as.matrix() %>% melt_dist() #making long data of distance matrices

#Adding sample type and treatment name. 
#this can be also done by merging metadata into the `horn_dist_long`
names <- data.frame(str_split_fixed(horn_dist_long$iso1, "_", 3))
names2 <- data.frame(str_split_fixed(horn_dist_long$iso2, "_", 3))
horn_dist_long$sample_id_1 <- paste(names$X1, names$X2, sep = "_")
horn_dist_long$method_1 <- ifelse(grepl("lyPMA", horn_dist_long$iso1),"lypma", 
                                         ifelse(grepl("ben", horn_dist_long$iso1),"benzonase", 
                                                ifelse(grepl("host", horn_dist_long$iso1),"host_zero", 
                                                       ifelse(grepl("qia", horn_dist_long$iso1),"qiaamp", 
                                                              ifelse(grepl("moly", horn_dist_long$iso1),"molysis", 
                                                                     "control")))))


#Adding data for iso 2 also should be done
horn_dist_long$sample_id_2 <- paste(names2$X1, names2$X2, sep = "_")
horn_dist_long$method_2 <-ifelse(grepl("lyPMA", horn_dist_long$iso2),"lypma", 
                                        ifelse(grepl("ben", horn_dist_long$iso2),"benzonase", 
                                               ifelse(grepl("host", horn_dist_long$iso2),"host_zero", 
                                                      ifelse(grepl("qia", horn_dist_long$iso2),"qiaamp", 
                                                             ifelse(grepl("moly", horn_dist_long$iso2),"molysis", 
                                                                    "control")))))


#subsetting distances of my interest
horn_dist_long$sample_id_1 <- ifelse(grepl("pos", horn_dist_long$sample_id_1, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", horn_dist_long$sample_id_1, ignore.case = T),"Neg.",
                                        horn_dist_long$sample_id_1))
horn_dist_long$sample_id_2 <- ifelse(grepl("pos", horn_dist_long$sample_id_2, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", horn_dist_long$sample_id_2, ignore.case = T),"Neg.",
                                        horn_dist_long$sample_id_2))


path_horn_dist_long_within_sampleid_from_control <- subset(horn_dist_long, horn_dist_long$sample_id_1 == horn_dist_long$sample_id_2) # data within samples

path_horn_dist_long_within_sampleid_from_control <- subset(path_horn_dist_long_within_sampleid_from_control,
                                                           path_horn_dist_long_within_sampleid_from_control$method_1 != path_horn_dist_long_within_sampleid_from_control$method_2) # remove irrelevant association

path_horn_dist_long_within_sampleid_from_control <- subset(path_horn_dist_long_within_sampleid_from_control, (path_horn_dist_long_within_sampleid_from_control$method_1 == "control") + (path_horn_dist_long_within_sampleid_from_control$method_2 == "control") != 0)


path_horn_dist_long_within_sampleid_from_control$treatment <- path_horn_dist_long_within_sampleid_from_control$method_1

path_horn_dist_long_within_sampleid_from_control$treatment <- ifelse(path_horn_dist_long_within_sampleid_from_control$treatment == "control", path_horn_dist_long_within_sampleid_from_control$method_2, path_horn_dist_long_within_sampleid_from_control$treatment) 


#Setting key method
path_horn_dist_long_within_sampleid_from_control$sample_type <- ifelse(grepl("NS", path_horn_dist_long_within_sampleid_from_control$iso1), "Nasal",
                                                                  ifelse(grepl("CFB", path_horn_dist_long_within_sampleid_from_control$iso1), "Sputum",
                                                                         ifelse(grepl("BAL", path_horn_dist_long_within_sampleid_from_control$iso1), "BAL",
                                                                                ifelse(grepl("pos|POS", path_horn_dist_long_within_sampleid_from_control$iso1, ignore.case = T), "Mock",
                                                                                       ifelse(grepl("neg|N_EXT", path_horn_dist_long_within_sampleid_from_control$iso1), "Neg.",NA)))))

#Making a column for baseline (controls, from where?)
path_horn_dist_long_within_sampleid_from_control <- path_horn_dist_long_within_sampleid_from_control %>% 
        mutate(dist_from = case_when(method_1 == "control" ~ iso1,
                                     method_2 == "control" ~ iso2))

dummy <- data.frame(iso1 = path_horn_dist_long_within_sampleid_from_control$dist_from %>% unique,
           iso2 = path_horn_dist_long_within_sampleid_from_control$dist_from %>% unique,
           dist = 0,
           treatment = "Untreated",
           method_1 = "control",
           method_2 = "control"
           )
names <- data.frame(str_split_fixed(dummy$iso1, "_", 3))
names2 <- data.frame(str_split_fixed(dummy$iso2, "_", 3))
dummy$sample_id_1 <- paste(names$X1, names$X2, sep = "_")
#Adding data for iso 2 also should be done
dummy$sample_id_2 <- paste(names2$X1, names2$X2, sep = "_")


#subsetting distances of my interest
dummy$sample_id_1 <- ifelse(grepl("pos", dummy$sample_id_1, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", dummy$sample_id_1, ignore.case = T),"Neg.",
                                        dummy$sample_id_1))
dummy$sample_id_2 <- ifelse(grepl("pos", dummy$sample_id_2, ignore.case = T),"Mock", 
                                 ifelse(grepl("neg|n_", dummy$sample_id_2, ignore.case = T),"Neg.",
                                        dummy$sample_id_2))
dummy$sample_type <- ifelse(grepl("NS", dummy$iso1), "Nasal",
                            ifelse(grepl("CFB", dummy$iso1), "Sputum",
                                   ifelse(grepl("BAL", dummy$iso1), "BAL",
                                          ifelse(grepl("pos|POS", dummy$iso1, ignore.case = T), "Mock",
                                                 ifelse(grepl("neg|N_EXT", dummy$iso1), "Neg.",NA)))))
dummy <- subset(dummy, !is.na(dummy$sample_type))
path_horn_dist_long_within_sampleid_from_control <- bind_rows(path_horn_dist_long_within_sampleid_from_control, dummy)


path_horn_dist_long_within_sampleid_from_control$subject_id <- path_horn_dist_long_within_sampleid_from_control$sample_id_1

path_horn_dist_long_within_sampleid_from_control$treatment <-
        factor(path_horn_dist_long_within_sampleid_from_control$treatment,
               levels = c("Untreated", "lypma", "benzonase", "host_zero", "molysis", "qiaamp"))


f4b2 <- path_horn_dist_long_within_sampleid_from_control %>% 
        mutate(across(sample_type, factor, levels=c(#"Mock", 
                                                    "BAL", "Nasal","Sputum"
                                                    ))) %>%
        subset(., .$sample_type != "Neg.") %>% 
        subset(., .$treatment != "Untreated") %>%
  mutate(treatment = factor(treatment, levels = c("lypma", "benzonase", "host_zero", "molysis", "qiaamp"),
                            labels = c("lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))) %>%
        ggplot(aes(x = dist, y = treatment, col = treatment)) +
        geom_boxplot() + 
        facet_wrap(~sample_type, nrow = 4) +
        scale_y_discrete(limits=rev) +
        scale_color_manual(values = c(#"#e31a1c",
                                      "#fb9a99","#33a02c",
                                      "#b2df8a","#1f78b4","#a6cee3"),
                           name = "Treatment",
                           breaks = c(#"Untreated", 
                                      "lyPMA", "Benzonase",
                                      "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        xlab("Morisita-Horn dissimilarity from untreated") +
        ylab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        theme(plot.tag = element_text(size = 15),
              axis.text.y = element_blank(),
              axis.ticks.y = element_blank(),
              legend.position = "none") +
        labs(tag = "B") +
        geom_vline(xintercept = 0, col = "black", linetype="dotted") +
        #coord_cartesian(xlim=c(-0.5, 1)) +
        #geom_text(aes(x = 0, label = treatment), hjust = 0, nudge_x = -.55, size = 3, color = "black", family = "sans") +
        #geom_text(aes(x = 0, label = text), hjust = 0, nudge_x = -0.4, size = 3, color = "black", family = "sans") +
        scale_x_continuous(breaks = c(-0.25, 0, 0.25, 0.5, 0.75),
                           labels = c(-0.25, "0 (low bias)", 0.25, 0.5, "0.75 (high bias)"))




figS8 <- ggarrange(f4a, f4b2, ncol = 1, common.legend = T, align = "hv") +
        guides(fill = guide_legend(nrow = 1))


figS8

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS10.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 170, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue
figS8

# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Functional richness - LMER all samples

Effect of some treatment was neutralized by interaction term. Therefore, the association was sample_type specific.

Effect size, standard error (SE) and p-value of a statistical test for functional richness with an interaction term using linear mixed effect model (Species richness ~ sample type * treatment + (1|subject_id) ).

raw result

library(lmerTest)
sample_data <- sample_data(phyloseq$phyloseq_path_rpk)
sample_data$log_centered_final_reads <- log(sample_data$Final_reads + 1) - median(log((subset(sample_data, sample_data$sample_type %in% c("BAL") & sample_data$treatment %in% c("Untreated")) %>% .$Final_reads) + 1))

lmer(S.obs ~ sample_type * treatment + (1|subject_id), data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum"))) 

## Linear mixed model fit by REML ['lmerModLmerTest']
## Formula: S.obs ~ sample_type * treatment + (1 | subject_id)
##    Data: sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL",  
##     "Nasal", "Sputum"))
## REML criterion at convergence: 855.5247
## Random effects:
##  Groups     Name        Std.Dev.
##  subject_id (Intercept) 35.90   
##  Residual               44.53   
## Number of obs: 95, groups:  subject_id, 20
## Fixed Effects:
##                          (Intercept)                      sample_typeNasal  
##                                23.60                                126.60  
##                    sample_typeSputum                        treatmentlyPMA  
##                               158.00                                 63.00  
##                   treatmentBenzonase                     treatmentHostZERO  
##                               137.20                                177.80  
##                     treatmentMolYsis                       treatmentQIAamp  
##                               203.20                                139.40  
##      sample_typeNasal:treatmentlyPMA      sample_typeSputum:treatmentlyPMA  
##                               -46.90                                 23.00  
##  sample_typeNasal:treatmentBenzonase  sample_typeSputum:treatmentBenzonase  
##                              -120.88                                -45.60  
##   sample_typeNasal:treatmentHostZERO   sample_typeSputum:treatmentHostZERO  
##                              -115.48                                -32.00  
##    sample_typeNasal:treatmentMolYsis    sample_typeSputum:treatmentMolYsis  
##                              -155.30                                -52.80  
##     sample_typeNasal:treatmentQIAamp     sample_typeSputum:treatmentQIAamp  
##                               -62.92                                -15.80

Tidy table

lmer(S.obs ~ sample_type * treatment + (1|subject_id), data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL", "Nasal", "Sputum"))) %>% 
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("sample_type|treatment", "", x))  %>% mutate(x = gsub(":", " * ", x)) %>% 
        column_to_rownames(var = "x") %>%  
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        mutate("Effect size (95% CI)" = paste(round(Estimate, 1) %>% format(nsmall = 1), 
                                " (", 
                                round(Estimate - 1.96 * abs(t_value), 1) %>% format(nsmall = 1),
                                ", ",
                                round(Estimate + 1.96 * abs(t_value), 1) %>% format(nsmall = 1),
                                ")", 
                                sep = ""),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>% 
        kbl(format = "html", escape = 0) %>% kable_styling(full_width = 0, html_font = "sans")

Table S9. Functional richness - all & stratified

Table S9. Effect size, standard error (SE) and p-value of a statistical test for functional richness with an interaction term using linear mixed effect model (functional richness ~ treatment + (1|subject_id)). Stratified analyses were conducted for each sample type as an interaction term of sample type and treatment was significant (p-value < 0.001) at ANOVA of LMER(functional richness ~ sample type + treatment + sample type * treatment + (1|subject_id)). The baseline of categorical variables is untreated group. Statistical significances were noted with : p-value < 0.01 and *: p-value < 0.001.

Association was not adjusted with sequencing depth.

Raw result - Mock

pfr_lmer_mock <- lm(S.obs ~ treatment, data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Mock"))) 
pfr_lmer_mock %>% summary

## 
## Call:
## lm(formula = S.obs ~ treatment, data = sample_data %>% data.frame %>% 
##     subset(., .$sample_type %in% c("Mock")))
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -59.600  -4.167  -1.000   3.667  61.600 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         434.333      9.269  46.858  < 2e-16 ***
## treatmentlyPMA      -42.733     13.748  -3.108  0.00465 ** 
## treatmentBenzonase  -86.933     13.748  -6.323 1.28e-06 ***
## treatmentHostZERO  -100.733     13.748  -7.327 1.12e-07 ***
## treatmentMolYsis    -14.133     13.748  -1.028  0.31380    
## treatmentQIAamp     -40.333     13.748  -2.934  0.00708 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 22.7 on 25 degrees of freedom
## Multiple R-squared:  0.7624, Adjusted R-squared:  0.7148 
## F-statistic: 16.04 on 5 and 25 DF,  p-value: 4.177e-07

Raw result - BAL

dummy1 <- ((phyloseq$phyloseq_path_rpk %>% otu_table %>% data.frame(check.names = F)) != 0) %>% colSums()


pfr_lmer_bal <- lm(S.obs ~ treatment, data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("BAL"))) 

pfr_lmer_bal %>% summary

## 
## Call:
## lm(formula = S.obs ~ treatment, data = sample_data %>% data.frame %>% 
##     subset(., .$sample_type %in% c("BAL")))
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -142.0  -49.3    0.2   60.3  158.4 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)           23.60      38.25   0.617 0.543098    
## treatmentlyPMA        63.00      54.10   1.165 0.255666    
## treatmentBenzonase   137.20      54.10   2.536 0.018135 *  
## treatmentHostZERO    177.80      54.10   3.286 0.003113 ** 
## treatmentMolYsis     203.20      54.10   3.756 0.000974 ***
## treatmentQIAamp      139.40      54.10   2.577 0.016553 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 85.54 on 24 degrees of freedom
## Multiple R-squared:  0.4487, Adjusted R-squared:  0.3338 
## F-statistic: 3.906 on 5 and 24 DF,  p-value: 0.009864

Raw result - BAL

pfr_lmer_ns <- lm(S.obs ~ treatment, data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Nasal"))) 
pfr_lmer_ns %>% summary

## 
## Call:
## lm(formula = S.obs ~ treatment, data = sample_data %>% data.frame %>% 
##     subset(., .$sample_type %in% c("Nasal")))
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
##  -57.6  -17.6    3.4   20.7   75.8 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         150.200      9.751  15.403 1.69e-15 ***
## treatmentlyPMA        7.800     16.890   0.462 0.647663    
## treatmentBenzonase   23.400     16.890   1.385 0.176484    
## treatmentHostZERO    69.400     16.890   4.109 0.000297 ***
## treatmentMolYsis     56.200     16.890   3.327 0.002391 ** 
## treatmentQIAamp      69.400     16.890   4.109 0.000297 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 30.84 on 29 degrees of freedom
## Multiple R-squared:  0.5248, Adjusted R-squared:  0.4428 
## F-statistic: 6.404 on 5 and 29 DF,  p-value: 0.0004024

Raw result - BAL

pfr_lmer_spt <- lm(S.obs ~ treatment, data = sample_data %>% data.frame %>% subset(., .$sample_type %in% c("Sputum"))) 
pfr_lmer_spt %>% summary

## 
## Call:
## lm(formula = S.obs ~ treatment, data = sample_data %>% data.frame %>% 
##     subset(., .$sample_type %in% c("Sputum")))
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -145.20  -18.00   -5.00   14.25  116.40 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)          181.60      22.76   7.978 3.32e-08 ***
## treatmentlyPMA        86.00      32.19   2.671 0.013353 *  
## treatmentBenzonase    91.60      32.19   2.845 0.008934 ** 
## treatmentHostZERO    145.80      32.19   4.529 0.000138 ***
## treatmentMolYsis     150.40      32.19   4.672 9.57e-05 ***
## treatmentQIAamp      123.60      32.19   3.839 0.000790 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 50.9 on 24 degrees of freedom
## Multiple R-squared:  0.5542, Adjusted R-squared:  0.4613 
## F-statistic: 5.967 on 5 and 24 DF,  p-value: 0.00101

Tidy table

pfr_lmer_mock_kbl <- pfr_lmer_mock %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              pfr_lmer_mock %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        

pfr_lmer_bal_kbl <- pfr_lmer_bal %>% 
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              pfr_lmer_bal %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
pfr_lmer_ns_kbl <- pfr_lmer_ns %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              pfr_lmer_ns %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
pfr_lmer_spt_kbl <-  pfr_lmer_spt %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              pfr_lmer_spt %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


tables9 <- cbind(pfr_lmer_bal_kbl, pfr_lmer_ns_kbl, pfr_lmer_spt_kbl) %>%
    kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 1, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")

tables9

save_kable(tables9, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS9.html", self_contained = T)

Function beta - all samples

Rwa result, with interaction term

phyloseq_rel_nz_f <- transform_sample_counts(phyloseq$phyloseq_path_rpk, function(x) {x/sum(x)}) %>%
        subset_samples(S.obs != 0 & sample_type %in% c("Mock", "BAL", "Nasal", "Sputum"))


set.seed(seed)
horn_perm_inter_fun <- vegan::adonis2(by = "terms", distance(phyloseq_rel_nz_f, method="horn") ~ sample_type * treatment + subject_id,
                                  data = phyloseq_rel_nz_f %>% sample_data %>% data.frame(check.names = F), 
                                  strata = phyloseq_rel_nz_f %>% sample_data %>% data.frame(check.names = F) %>% .$subject_id,
                                  permutations = 10000)
set.seed(seed)
horn_perm_ns_fun <- vegan::adonis2(by = "terms", distance(subset_samples(phyloseq_rel_nz_f, sample_type == "Nasal"), method="horn") ~ lypma + benzonase + host_zero + molysis + qiaamp,
                               data = subset_samples(phyloseq_rel_nz_f, sample_type == "Nasal") %>%
                                       sample_data %>% data.frame(check.names = F),
                               strata = subset_samples(phyloseq_rel_nz_f, sample_type == "Nasal") %>% 
                                       sample_data %>% data.frame(check.names = F) %>% .$subject_id, permutations = 10000)
set.seed(seed)
horn_perm_bal_fun  <- vegan::adonis2(by = "terms", distance(subset_samples(phyloseq_rel_nz_f, sample_type == "BAL"), method="horn") ~  lypma + benzonase + host_zero + molysis + qiaamp,
                                 data = subset_samples(phyloseq_rel_nz_f, sample_type == "BAL") %>% sample_data %>% data.frame(check.names = F),
                                 strata = subset_samples(phyloseq_rel_nz_f, sample_type == "BAL") %>%
                                         sample_data %>% data.frame(check.names = F) %>% .$subject_id,
                                  permutations = 10000)
set.seed(seed)
horn_perm_spt_fun <- vegan::adonis2(by = "terms", distance(subset_samples(phyloseq_rel_nz_f, sample_type == "Sputum"), method="horn") ~ lypma + benzonase + host_zero + molysis + qiaamp,
                                data = subset_samples(phyloseq_rel_nz_f, sample_type == "Sputum") %>% sample_data %>% data.frame(check.names = F),
                                strata = subset_samples(phyloseq_rel_nz_f, sample_type == "Sputum")
                                %>% sample_data %>% data.frame(check.names = F) %>% .$subject_id,
                                  permutations = 10000)


horn_perm_inter_fun

Tidy table of PERMANOVA result

horn_perm_inter_fun %>% data.frame(check.names = F) %>% rownames_to_column("row.names") %>% 
        mutate(row.names = case_when(row.names == "sample_type" ~ 'Sample type',
                                     row.names == "treatment" ~ 'Treatment',
                                     row.names == "subject_id" ~ 'Subject',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "sample_type:treatment" ~ 'Sample type * Treatment',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
        mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3),
               `Pr(>F)` = format(`Pr(>F)`, nsmall = 3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        dplyr::select(c("Degree of freedom", "R<sup>2</sup>", "<i>p</i>-value", " ")) %>% 
        kbl(format = "html", escape = 0) %>%
        kable_styling(full_width = 0, html_font = "sans")

Function beta - stratified

Not included in the main text

Table. Degree of freedom, effect size (residual, R^2) and p-value of permutational ANOVA for functional Horn-Morisita distances with an interaction term and strata term (BC-distance of functions ~ sample type * treatment + log10(final reads), strata = subject id).

Raw result - BAL

horn_perm_bal_fun

Raw result - Nasal

horn_perm_ns_fun

Raw result - Sputum

horn_perm_spt_fun

Tidy table

a <- horn_perm_bal_fun %>% data.frame(check.names = F) %>% rownames_to_column('row.names') %>% 
        mutate(row.names = case_when(row.names == "lypma" ~ 'lyPMA',
                                     row.names == "benzonase" ~ 'Benzonase',
                                     row.names == "host_zero" ~ 'HostZERO',
                                     row.names == "molysis" ~ 'MolYsis',
                                     row.names == "qiaamp" ~ 'QIAamp',
                                     row.names == "subject_id" ~ 'Subject id',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
                mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        dplyr::select(c("R<sup>2</sup>", "<i>p</i>-value", " ")) 

b <- horn_perm_ns_fun %>% data.frame(check.names = F) %>% rownames_to_column('row.names') %>% 
        mutate(row.names = case_when(row.names == "lypma" ~ 'lyPMA',
                                     row.names == "benzonase" ~ 'Benzonase',
                                     row.names == "host_zero" ~ 'HostZERO',
                                     row.names == "molysis" ~ 'MolYsis',
                                     row.names == "qiaamp" ~ 'QIAamp',
                                     row.names == "subject_id" ~ 'Subject id',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
                mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        dplyr::select(c("R<sup>2</sup>", "<i>p</i>-value", " ")) 

c <- horn_perm_spt_fun %>% data.frame(check.names = F) %>% rownames_to_column('row.names') %>% 
        mutate(row.names = case_when(row.names == "lypma" ~ 'lyPMA',
                                     row.names == "benzonase" ~ 'Benzonase',
                                     row.names == "host_zero" ~ 'HostZERO',
                                     row.names == "molysis" ~ 'MolYsis',
                                     row.names == "qiaamp" ~ 'QIAamp',
                                     row.names == "subject_id" ~ 'Subject id',
                                     row.names == "log10(Final_reads)" ~ 'log10(Final reads)',
                                     row.names == "Residual" ~ 'Residual',
                                     row.names == "Total" ~ 'Total')) %>% column_to_rownames('row.names') %>% 
                mutate(` ` = case_when(abs(`Pr(>F)`) < 0.001 ~ "***",
                                            abs(`Pr(>F)`) < 0.01 ~ "**",
                                            abs(`Pr(>F)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        mutate(across(is.numeric, round, digits=3)) %>% 
        rename("<i>p</i>-value" = "Pr(>F)",
               "R<sup>2</sup>" = "R2",
               "Degree of freedom" = "Df") %>% 
        dplyr::select(c("R<sup>2</sup>", "<i>p</i>-value", " ")) 


tables9_A <- cbind(a, b, c) %>% 
        kbl(format = "html", escape = 0) %>%
        add_header_above(c(" " = 1, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        kable_styling(full_width = 0, html_font = "sans")


tables9_A

MaAslin renaming

rename_function_gruop <- function(list_maaslin_result){
        taxa_df <- tax_table(phyloseq$phyloseq_path_cpm) %>% 
                data.frame %>% remove_rownames() %>% rename(feature = "pathway")
        tax_table(phyloseq$phyloseq_path_cpm)
        list_maaslin_result$data <-
                list_maaslin_result$data %>% 
                merge(., taxa_df, by = "feature") %>% 
                dplyr::select(-c("feature")) %>%
                rename(feature = "group") %>%
                dplyr::select(c("feature", "lypma", "benzonase", "host_zero", "molysis", "qiaamp"))
        list_maaslin_result$data_italic <- 
                list_maaslin_result$data_italic %>% 
                rownames_to_column("feature") %>%
                merge(., taxa_df, by = "feature") %>% 
                dplyr::select(-c("feature")) %>%
                column_to_rownames("group")
        list_maaslin_result$data_sig <-
                list_maaslin_result$data_sig  %>% 
                        rownames_to_column("feature") %>%
                merge(., taxa_df, by = "feature") %>% 
                dplyr::select(-c("feature")) %>%
                column_to_rownames("group")
        list_maaslin_result$rel <-
                list_maaslin_result$rel %>% 
                merge(., taxa_df, by = "feature") %>% 
                dplyr::select(-c("feature")) %>%
                rename(feature = "group") %>%
                dplyr::select(c("feature", "Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"))
        list_maaslin_result
        
}


#f_fit_data_pos <- rename_function_gruop(f_fit_data_pos)
f_fit_data_bal <- rename_function_gruop(f_fit_data_bal)
f_fit_data_ns <- rename_function_gruop(f_fit_data_ns)
f_fit_data_spt <- rename_function_gruop(f_fit_data_spt)

Functional MaAsLin raw results

f_maaslin_all

Fig. S10 MaAslin function - volcano plot

Fig. S10. Volcano plot of differential abundance of function by each treatment with a model MaAsLin (copies per million of each function ~ sample type + lyPMA + Benzonase + HostZERO + MolYsis + QIAamp, random effect = subject id).

#Volcano plot

figS9 <- ggplot(f_maaslin_all, aes(y = -log10(qval), x = Estimate, col = metadata)) +
        theme_classic(base_family = "sans") +
        #labs(tag = "A") +
        geom_point(size = 2, alpha = 0.3, stroke = 0) +
        xlab("Change estimate of CLR-normalized CPM") +
        ylab("-log<sub>10</sub>(*q*-value)") +
        geom_hline(yintercept = 1, col = "gray") +
        geom_vline(xintercept = 0, col = "gray") +
        annotate(family = "sans",
                 geom='richtext',
                 x=0, y=10,
                 label = "<i>q</i>-value = 0.1, fold-change = 0") +
        theme(legend.position = "top", axis.title.y = ggtext::element_markdown()) +
        scale_color_manual(values = c("#a65628",
                                      "grey",
                                      "#fb9a99",
                                      "#33a02c",
                                      "#b2df8a",
                                      "#1f78b4",
                                      "#a6cee3"),
                           breaks = c("log10.Final_reads",
                                      "sample_type",
                                      "lypma",
                                      "benzonase",
                                      "host_zero",
                                      "molysis",
                                      "qiaamp"), 
                           labels = c("log<sub>10</sub>(Final reads)",
                                      "Sample type",
                                      "lyPMA",
                                      "Benzonase",
                                      "HostZERO",
                                      "MolYsis",
                                      "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        guides(col = guide_legend(title = "Factors", title.position = "top", nrow = 2))


png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS11.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 90, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

figS9
# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

figS9

Fig. S11 Function baloon plot

Fig. S11. Mean copies per million of top 20 significant function identified by differential abundance analysis using MaAsLin. Analyses were stratified by sample type. (A) Mock community, (B) bronchoalveolar lavage, (C) nasal swabs, and (D) sputum. Statistical significances were noted at the level of q-value < 0.1.

#ffff33 qia

f5b <- merge(f_fit_data_bal$rel %>%
              gather(treatment,
                     value,
                     Untreated:QIAamp,
                     factor_key=TRUE),
      f_fit_data_bal$data_italic %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     qval,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
      by.x = c('feature', 'treatment'),
      by.y = c('feature', 'treatment'),
      all = T) %>%
        
        merge(f_fit_data_bal$data_sig %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     sig,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
              by.x = c('feature', 'treatment'),
              by.y = c('feature', 'treatment'),
              all = T) %>%
        mutate(sig = case_when(sig < 0.1 ~ "< 0.1",
                               .default = "> 0.1"),
               value = value * 1000000) %>%



#Baloon plot
        ggballoonplot(size = "value", y = "feature", x= "treatment", fill = "sig") +
        theme_classic(base_family = "sans") +
        #colors for qvalues
        #xlab("Experimental group") +
        #ylab("Species") +
        #labs(tag = "A")  +
        ggtitle("A  BAL") +
        theme(panel.grid.major = element_line(colour = "grey"),
              legend.position = "top",
              axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
              #Element markdown for taxa name italicizing
              axis.text.y = ggtext::element_markdown(size = 8),
              axis.title.y = element_blank(),
              axis.title.x = element_blank(),
              plot.margin = unit(c(0,0.2,0,1), 'lines'))  +
        scale_fill_manual(values = c("red", "grey"), aes(y = feature,
                      x = treatment,
                      label = sig)) +
        guides(fill = guide_legend(title = c(expression(paste(italic("q"),
                                                       "-value",
                                                       sep = ""))),
                                   title.position = "top",
                                   nrow = 2,
                                   override.aes = list(size=3)),
               size = guide_legend(title = "Copies per million",
                                   title.position = "top",
                                   order = 1,
                                   nrow = 1)
               ) + 
        scale_x_discrete(labels=c("control" = "Untreated",
                                  "lypma" = "lyPMA",
                                  "benzonase" = "Benzonase",
                                  "host_zero" = "Host-zero",
                                  "molysis" = "MolYsis",
                                  "qiaamp" = "QIAamp")
                         )

f5c <- merge(f_fit_data_ns$rel %>%
              gather(treatment,
                     value,
                     Untreated:QIAamp,
                     factor_key=TRUE),
      f_fit_data_ns$data_italic %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     qval,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
      by.x = c('feature', 'treatment'),
      by.y = c('feature', 'treatment'),
      all = T) %>%
        
        merge(f_fit_data_ns$data_sig %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     sig,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
              by.x = c('feature', 'treatment'),
              by.y = c('feature', 'treatment'),
              all = T) %>%
        mutate(sig = case_when(sig < 0.1 ~ "< 0.1",
                               .default = "> 0.1"),
               value = value * 1000000) %>%


#Baloon plot
        ggballoonplot(size = "value", y = "feature", x= "treatment", fill = "sig") +
        theme_classic(base_family = "sans") +
        #colors for qvalues
        #xlab("Experimental group") +
        #ylab("Species") +
        #labs(tag = "B")  +
        ggtitle("B  Nasal swab") +
        theme(panel.grid.major = element_line(colour = "grey"),
              legend.position = "top",
              axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
              #Element markdown for taxa name italicizing
              axis.text.y = ggtext::element_markdown(size = 8),
              axis.title.y = element_blank(),
              axis.title.x = element_blank(),
              plot.margin = unit(c(0,0.2,0,1), 'lines'))  +
        scale_fill_manual(values = c("red", "grey"), aes(y = feature,
                      x = treatment,
                      label = sig)) +
        guides(fill = guide_legend(title = c(expression(paste(italic("q"),
                                                       "-value",
                                                       sep = ""))),
                                   title.position = "top",
                                   override.aes = list(size=3)),
               size = guide_legend(title = "Copies per million",
                                   title.position = "top",
                                   order = 1,
                                   nrow = 1)
               ) + 
        scale_x_discrete(labels=c("control" = "Untreated",
                                  "lypma" = "lyPMA",
                                  "benzonase" = "Benzonase",
                                  "host_zero" = "Host-zero",
                                  "molysis" = "MolYsis",
                                  "qiaamp" = "QIAamp")
                         )

f5d <- merge(f_fit_data_spt$rel %>%
              gather(treatment,
                     value,
                     Untreated:QIAamp,
                     factor_key=TRUE),
      f_fit_data_spt$data_italic %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     qval,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
      by.x = c('feature', 'treatment'),
      by.y = c('feature', 'treatment'),
      all = T) %>%
        merge(f_fit_data_spt$data_sig %>%
              rownames_to_column("feature") %>%
              gather(treatment,
                     sig,
                     lyPMA:QIAamp,
                     factor_key=TRUE),
              by.x = c('feature', 'treatment'),
              by.y = c('feature', 'treatment'),
              all = T) %>%
        mutate(sig = case_when(sig < 0.1 ~ "< 0.1",
                               .default = "> 0.1"),
               value = value * 1000000) %>%


#Baloon plot
        ggballoonplot(size = "value", y = "feature", x= "treatment", fill = "sig") +
        theme_classic(base_family = "sans") +
        #colors for qvalues
        #xlab("Experimental group") +
        #ylab("Species") +
        #labs(tag = "C")  +
        ggtitle("C  Sputum") +
        theme(panel.grid.major = element_line(colour = "grey"),
              legend.position = "top",
              axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
              #Element markdown for taxa name italicizing
              axis.text.y = ggtext::element_markdown(size = 8),
              axis.title.y = element_blank(),
              axis.title.x = element_blank(),
              plot.margin = unit(c(0,0.2,0,1), 'lines'))  +
        scale_fill_manual(values = c("red", "grey"), aes(y = feature,
                      x = treatment,
                      label = sig)) +
        guides(fill = guide_legend(title = c(expression(paste(italic("q"),
                                                       "-value",
                                                       sep = ""))),
                                   title.position = "top",
                                   override.aes = list(size=5)),
               size = guide_legend(title = "Copies per million",
                                   title.position = "top",
                                   order = 1,
                                   nrow = 1)
               ) + 
        scale_x_discrete(labels=c("control" = "Untreated",
                                  "lypma" = "lyPMA",
                                  "benzonase" = "Benzonase",
                                  "host_zero" = "Host-zero",
                                  "molysis" = "MolYsis",
                                  "qiaamp" = "QIAamp")
                         )

figS10 <- ggarrange(f5d %>% lemon::g_legend() %>% as_ggplot,
                  f5b,
                  f5c,
                  f5d,
                  ncol=1, heights = c(1.5, 4, 4, 4),
                  legend = "none",
                  align = "hv")


annotate_figure(figS10,
                left = text_grob("Predicted function",
                                 rot = 90,
                                 family = "sans", 
                                 size = 11),
                bottom = text_grob("Treatment",
                                 rot = 0,
                                 family = "sans", 
                                 size = 11, hjust = -3)
                
)

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS12.png",   # The directory you want to save the file in
    width = 240, # The width of the plot in inches
    height = 220, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue



annotate_figure(figS10,
                left = text_grob("Predicted function",
                                 rot = 90,
                                 family = "sans", 
                                 size = 11),
                bottom = text_grob("Treatment",
                                 rot = 0,
                                 family = "sans", 
                                 size = 11, hjust = -3)
                
)


# alpha diversity plots
#ggarrange(f4ad, ggarrange(f4e, f4f, ncol = 2),
#          ncol = 1) # alpha diversity plots

dev.off()

## quartz_off_screen 
##                 2

Fig. S12. Rarefraction curve of species richenss

Fig. S12. Rarefaction curve for (A) species richness and (B) function richenss stratified by sample type, after removing possible contaminant-taxa identified by decontam and low prevalent taxa.

As a sanity check, rarefaction curves were generated and seemed to be saturated

fig_rarefraction <- phyloseq$phyloseq_count %>% 
        sample_data %>% 
        data.frame %>%
        subset(., !is.na(.$treatment) & sample_type %in% c("BAL", "Nasal", "Sputum")) %>%
ggplot(., aes(x = log10(Final_reads/1000000), y = S.obs, col = treatment)) +
        geom_point() +
        theme_classic(base_family = "sans") +
        #xlab("Final reads x 10<sup>6</sup>") +
        ylab("Microbial species richness") +
        labs(tag = "A") +
        theme(axis.title.x = element_blank(), legend.position = "top") +
        guides(col = guide_legend(title = "Treatment", title.position = "top", nrow = 1)) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                          name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        facet_wrap(~sample_type, scales = "free", nrow = 1)

fig_rarefraction_function <- phyloseq$phyloseq_path_rpk %>% 
        sample_data %>% 
        data.frame %>%
        subset(., !is.na(.$treatment) & sample_type %in% c("BAL", "Nasal", "Sputum")) %>%
ggplot(., aes(x = log10(Final_reads/1000000), y = S.obs, col = treatment)) +
        geom_point() +
        theme_classic(base_family = "sans") +
        xlab("log<sub>10</sub>(Final reads x 10<sup>6</sup>)") +
        ylab("Functional richness") +
        labs(tag = "C") +
        theme(axis.title.x = element_markdown(), legend.position = "top") +
        guides(col = guide_legend(title = "Treatment", title.position = "top", nrow = 1)) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                          name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        facet_wrap(~sample_type, scales = "free", nrow = 1)



fig_rarefraction_viral <- phyloseq$phyloseq_count %>% 
        sample_data %>% 
        data.frame %>%
        subset(., !is.na(.$treatment) & sample_type %in% c("BAL", "Nasal", "Sputum")) %>%
ggplot(., aes(x = log10(Final_reads/1000000), y = V.obs, col = treatment)) +
        geom_point() +
        theme_classic(base_family = "sans") +
        #xlab("Final reads x 10<sup>6</sup>") +
        ylab("Viral richness") +
        labs(tag = "B") +
        theme(axis.title.x = element_blank(), legend.position = "top",) +
        guides(col = guide_legend(title = "Treatment", title.position = "top", nrow = 1)) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                          name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        facet_wrap(~sample_type, scales = "free", nrow = 1)



png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS13_log.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 180, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

ggarrange(fig_rarefraction, fig_rarefraction_viral, fig_rarefraction_function, common.legend = T, ncol = 1)

dev.off()

## quartz_off_screen 
##                 2

fig_rarefraction <- phyloseq$phyloseq_count %>% 
        sample_data %>% 
        data.frame %>%
        subset(., !is.na(.$treatment) & sample_type %in% c("BAL", "Nasal", "Sputum")) %>%
ggplot(., aes(x = Final_reads/1000000, y = S.obs, col = treatment)) +
        geom_point() +
        theme_classic(base_family = "sans") +
        #xlab("Final reads x 10<sup>6</sup>") +
        ylab("Microbial species richness") +
        labs(tag = "A") +
        theme(axis.title.x = element_blank(), legend.position = "top") +
        guides(col = guide_legend(title = "Treatment", title.position = "top", nrow = 1)) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                          name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        facet_wrap(~sample_type, scales = "free", nrow = 1)

fig_rarefraction_function <- phyloseq$phyloseq_path_rpk %>% 
        sample_data %>% 
        data.frame %>%
        subset(., !is.na(.$treatment) & sample_type %in% c("BAL", "Nasal", "Sputum")) %>%
ggplot(., aes(x = Final_reads/1000000, y = S.obs, col = treatment)) +
        geom_point() +
        theme_classic(base_family = "sans") +
        xlab("Final reads x 10<sup>6</sup>") +
        ylab("Functional richness") +
        labs(tag = "C") +
        theme(axis.title.x = element_markdown(), legend.position = "top") +
        guides(col = guide_legend(title = "Treatment", title.position = "top", nrow = 1)) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                          name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        facet_wrap(~sample_type, scales = "free", nrow = 1)



fig_rarefraction_viral <- phyloseq$phyloseq_count %>% 
        sample_data %>% 
        data.frame %>%
        subset(., !is.na(.$treatment) & sample_type %in% c("BAL", "Nasal", "Sputum")) %>%
ggplot(., aes(x = Final_reads/1000000, y = V.obs, col = treatment)) +
        geom_point() +
        theme_classic(base_family = "sans") +
        #xlab("Final reads x 10<sup>6</sup>") +
        ylab("Viral richness") +
        labs(tag = "B") +
        theme(axis.title.x = element_blank(), legend.position = "top",) +
        guides(col = guide_legend(title = "Treatment", title.position = "top", nrow = 1)) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                          name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        facet_wrap(~sample_type, scales = "free", nrow = 1)

ggarrange(fig_rarefraction, fig_rarefraction_viral, fig_rarefraction_function, common.legend = T, ncol = 1)

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS13_updated.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 180, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

ggarrange(fig_rarefraction, fig_rarefraction_viral, fig_rarefraction_function, common.legend = T, ncol = 1)

dev.off()

## quartz_off_screen 
##                 2

Fig. S13. Species richness of decontaminated output

As some of the species richness got way higher, this data cannot be used for alpha diversity indices.

Fig. S13. Species richness of (A) raw data after prevalence and abundance filtering, (B) decontaminated species richness with decontam37, and (C) decontaminated data using tinyvamp.

f10a <- ggplot(subset(sample_data(phyloseq$phyloseq_count) %>% 
                             data.frame, sample_data(phyloseq$phyloseq_count)$sample_type %in% c("Sputum", "Nasal", "BAL"#, "Mock"
                                                                                                 )), aes(x = treatment, y = S.obs)) +
        geom_jitter(aes(color = treatment), position = position_jitter(0.2),
                    size = 1.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        ylab("Species richness") +
        xlab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using 
        labs(tag = "A") +
        ggtitle("Prevalence & abundance filtered data") +
        theme(plot.tag = element_text(size = 15),
              axis.text.x = element_blank(),
              axis.title.x = element_blank(),
              axis.title.y = element_blank(),
              axis.ticks.x = element_blank(),
              legend.position = "top") +
        facet_wrap(~sample_type, nrow = 1) + 
        guides(col = guide_legend(nrow = 1)) 

f10b <- ggplot(subset(sample_data(phyloseq$phyloseq_count) %>% 
                             data.frame, sample_data(phyloseq$phyloseq_count)$sample_type %in% c("Sputum", "Nasal", "BAL"#, "Mock"
                                                                                                 )), aes(x = treatment, y = S.obs)) +
        geom_jitter(aes(color = treatment), position = position_jitter(0.2),
                    size = 1.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        ylab("Species richness") +
        xlab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using 
        labs(tag = "B") +
        ggtitle("Decontaminated data 1") +
        theme(plot.tag = element_text(size = 15),
              axis.text.x = element_blank(),
              axis.title.x = element_blank(),
              axis.title.y = element_blank(),
              axis.ticks.x = element_blank(),
              legend.position = "top") +
        facet_wrap(~sample_type, nrow = 1) + 
        guides(col = guide_legend(nrow = 1)) 


f10c <- ggplot(subset(sample_data(phyloseq_tv) %>% 
                             data.frame, sample_data(phyloseq_tv)$sample_type %in% c("Sputum", "Nasal", "BAL", "Mock")), aes(x = treatment, y = S.obs)) +
        geom_jitter(aes(color = treatment), position = position_jitter(0.2),
                    size = 1.2, alpha = 0.3, stroke = 0) +
        stat_summary(aes(color = treatment),
                             fun.data="mean_sdl",  fun.args = list(mult=1), 
                             geom = "pointrange",  size = 0.4) +
        ylab("Species richness") +
        xlab("Treatment group") +
        theme_classic (base_size = 12, base_family = "sans") + 
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"), name = "Treatment", labels = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using 
        labs(tag = "C") +
        ggtitle("Decontaminated data 2") +
        theme(plot.tag = element_text(size = 15),
              axis.text.x = element_blank(), 
              axis.title.x = element_blank(),
              axis.title.y = element_blank(),
              axis.ticks.x = element_blank(),
              legend.position = "top") +
        facet_wrap(~sample_type, nrow = 1) + 
        guides(col = guide_legend(nrow = 1)) 

figa1 <- ggarrange(f10a, f10b, f10c, common.legend = T, ncol = 1)



annotate_figure(figa1,
                left = text_grob("Species richness",
                                 rot = 90,
                                 family = "sans", 
                                 size = 11),
                bottom = text_grob("Treatment",
                                 rot = 0,
                                 family = "sans", 
                                 size = 11)
)

png(file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/FigureS14.png",   # The directory you want to save the file in
    width = 180, # The width of the plot in inches
    height = 180, # The height of the plot in inches
    units = "mm",
    res = 600
) #fixing multiple page issue

annotate_figure(figa1,
                left = text_grob("Species richness",
                                 rot = 90,
                                 family = "sans", 
                                 size = 11),
                bottom = text_grob("Treatment",
                                 rot = 0,
                                 family = "sans", 
                                 size = 11)
)


dev.off()

## quartz_off_screen 
##                 2

Table S11. Species richness change after decontamination

Table S11. Effect size (95% confidence interval) and p-value of decontaminated species richness using decontam37 (decontaminated data 1) and tinyvamp (decontaminated data 2). The change was tested using a model lmer(species richness~ treatment + (1|subject id)). Statistical significances were noted with : p-value < 0.05, : p-value < 0.01, and : p-value < 0.001.

Raw results - decontam/BAL

sr_lmer_bal_decontam <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data(phyloseq_decontam$phyloseq_rel) %>% 
                             data.frame %>% 
                            subset(.,.$sample_type == "BAL"))
sr_lmer_bal_decontam %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: sample_data(phyloseq_decontam$phyloseq_rel) %>% data.frame %>%  
##     subset(., .$sample_type == "BAL")
## 
## REML criterion at convergence: 190.8
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.3988 -0.5046 -0.1313  0.5047  2.6286 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 93.81    9.685   
##  Residual               78.16    8.841   
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error     df t value Pr(>|t|)   
## (Intercept)           3.400      5.865  9.647   0.580  0.57537   
## treatmentlyPMA        0.600      5.591 20.000   0.107  0.91561   
## treatmentBenzonase    5.800      5.591 20.000   1.037  0.31197   
## treatmentHostZERO     8.200      5.591 20.000   1.467  0.15805   
## treatmentMolYsis     17.800      5.591 20.000   3.183  0.00467 **
## treatmentQIAamp       9.200      5.591 20.000   1.645  0.11552   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.477                            
## trtmntBnzns -0.477  0.500                     
## trtmntHZERO -0.477  0.500  0.500              
## trtmntMlYss -0.477  0.500  0.500  0.500       
## trtmntQIAmp -0.477  0.500  0.500  0.500  0.500

Raw results - decontam/nasal

sr_lmer_ns_decontam <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data(phyloseq_decontam$phyloseq_rel) %>% 
                             data.frame %>% 
                            subset(.,.$sample_type == "Nasal"))
sr_lmer_ns_decontam %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: sample_data(phyloseq_decontam$phyloseq_rel) %>% data.frame %>%  
##     subset(., .$sample_type == "Nasal")
## 
## REML criterion at convergence: 175.7
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.8245 -0.5693 -0.1774  0.6449  3.0756 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept)  1.089   1.043   
##  Residual               16.499   4.062   
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          8.4000     1.3262 28.7973   6.334  6.6e-07 ***
## treatmentlyPMA      -5.6016     2.2464 25.2088  -2.494 0.019560 *  
## treatmentBenzonase  -0.5743     2.2468 25.3111  -0.256 0.800319    
## treatmentHostZERO    8.4257     2.2468 25.3111   3.750 0.000924 ***
## treatmentMolYsis     5.2016     2.2464 25.2088   2.316 0.029002 *  
## treatmentQIAamp      6.3743     2.2468 25.3111   2.837 0.008837 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.554                            
## trtmntBnzns -0.554  0.315                     
## trtmntHZERO -0.554  0.315  0.346              
## trtmntMlYss -0.554  0.308  0.339  0.339       
## trtmntQIAmp -0.554  0.339  0.307  0.307  0.315

Raw results - decontam/spt

sr_lmer_spt_decontam <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data(phyloseq_decontam$phyloseq_rel) %>% 
                             data.frame %>% 
                            subset(.,.$sample_type == "Sputum"))
sr_lmer_spt_decontam %>% summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: sample_data(phyloseq_decontam$phyloseq_rel) %>% data.frame %>%  
##     subset(., .$sample_type == "Sputum")
## 
## REML criterion at convergence: 216.5
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -2.0585 -0.4479 -0.1131  0.3572  1.7178 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 300.6    17.34   
##  Residual               224.7    14.99   
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error      df t value Pr(>|t|)    
## (Intercept)          15.600     10.250   9.100   1.522  0.16198    
## treatmentlyPMA       35.000      9.481  20.000   3.692  0.00144 ** 
## treatmentBenzonase   63.400      9.481  20.000   6.687 1.65e-06 ***
## treatmentHostZERO    97.600      9.481  20.000  10.295 1.94e-09 ***
## treatmentMolYsis    107.000      9.481  20.000  11.286 3.99e-10 ***
## treatmentQIAamp      80.800      9.481  20.000   8.523 4.33e-08 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.462                            
## trtmntBnzns -0.462  0.500                     
## trtmntHZERO -0.462  0.500  0.500              
## trtmntMlYss -0.462  0.500  0.500  0.500       
## trtmntQIAmp -0.462  0.500  0.500  0.500  0.500

Raw results - Tinyvamp/BAL

sr_lmer_bal_tv <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data(phyloseq_tv) %>% 
                             data.frame %>% 
                            subset(.,.$sample_type == "BAL"))
sr_lmer_bal_tv %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data(phyloseq_tv) %>% data.frame %>% subset(., .$sample_type ==  
##     "BAL")
## 
## REML criterion at convergence: 162.6
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.60044 -0.62449 -0.00833  0.40102  2.22867 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 42.84    6.545   
##  Residual               44.84    6.697   
## Number of obs: 28, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error     df t value Pr(>|t|)  
## (Intercept)           3.619      4.497 12.920   0.805   0.4355  
## treatmentlyPMA        1.000      4.735 18.131   0.211   0.8351  
## treatmentBenzonase    3.981      4.541 18.318   0.877   0.3920  
## treatmentHostZERO     5.981      4.541 18.318   1.317   0.2041  
## treatmentMolYsis     11.981      4.541 18.318   2.638   0.0165 *
## treatmentQIAamp       6.181      4.541 18.318   1.361   0.1900  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.526                            
## trtmntBnzns -0.571  0.521                     
## trtmntHZERO -0.571  0.521  0.565              
## trtmntMlYss -0.571  0.521  0.565  0.565       
## trtmntQIAmp -0.571  0.521  0.565  0.565  0.565

Raw results - Tinyvamp/Nasal

sr_lmer_ns_tv <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data(phyloseq_tv) %>% 
                             data.frame %>% 
                            subset(.,.$sample_type == "Nasal")) 
sr_lmer_ns_tv %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data(phyloseq_tv) %>% data.frame %>% subset(., .$sample_type ==  
##     "Nasal")
## 
## REML criterion at convergence: 141.5
## 
## Scaled residuals: 
##      Min       1Q   Median       3Q      Max 
## -1.94756 -0.56006 -0.08626  0.42344  2.24527 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 0.3913   0.6255  
##  Residual               5.0338   2.2436  
## Number of obs: 35, groups:  subject_id, 10
## 
## Fixed effects:
##                    Estimate Std. Error       df t value Pr(>|t|)    
## (Intercept)         6.30000    0.73655 28.73600   8.553 2.17e-09 ***
## treatmentlyPMA     -1.51165    1.24265 25.29799  -1.216   0.2350    
## treatmentBenzonase  0.09912    1.24293 25.41039   0.080   0.9371    
## treatmentHostZERO   6.29912    1.24293 25.41039   5.068 2.99e-05 ***
## treatmentMolYsis    3.11165    1.24265 25.29799   2.504   0.0191 *  
## treatmentQIAamp     6.30088    1.24293 25.41039   5.069 2.98e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.550                            
## trtmntBnzns -0.550  0.312                     
## trtmntHZERO -0.550  0.312  0.348              
## trtmntMlYss -0.550  0.304  0.340  0.340       
## trtmntQIAmp -0.550  0.340  0.303  0.303  0.312

Raw results - Tinyvamp/Sputum

sr_lmer_spt_tv <- lmer(S.obs ~ treatment + (1|subject_id),
                    data = sample_data(phyloseq_tv) %>% 
                             data.frame %>% 
                            subset(.,.$sample_type == "Sputum")) 
sr_lmer_spt_tv %>% summary()

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: S.obs ~ treatment + (1 | subject_id)
##    Data: 
## sample_data(phyloseq_tv) %>% data.frame %>% subset(., .$sample_type ==  
##     "Sputum")
## 
## REML criterion at convergence: 183.2
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.6411 -0.5215 -0.1671  0.4478  1.9234 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  subject_id (Intercept) 75.09    8.665   
##  Residual               55.91    7.477   
## Number of obs: 30, groups:  subject_id, 5
## 
## Fixed effects:
##                    Estimate Std. Error     df t value Pr(>|t|)    
## (Intercept)           8.000      5.119  9.081   1.563  0.15221    
## treatmentlyPMA       14.800      4.729 20.000   3.130  0.00528 ** 
## treatmentBenzonase   26.200      4.729 20.000   5.540 2.01e-05 ***
## treatmentHostZERO    45.200      4.729 20.000   9.558 6.73e-09 ***
## treatmentMolYsis     51.800      4.729 20.000  10.954 6.70e-10 ***
## treatmentQIAamp      35.200      4.729 20.000   7.443 3.49e-07 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) trtPMA trtmnB tHZERO trtmMY
## trtmntlyPMA -0.462                            
## trtmntBnzns -0.462  0.500                     
## trtmntHZERO -0.462  0.500  0.500              
## trtmntMlYss -0.462  0.500  0.500  0.500       
## trtmntQIAmp -0.462  0.500  0.500  0.500  0.500

Tidy table of LMER for decontaminated species richness

sr_lmer_bal_decontam_kbl <- sr_lmer_bal_decontam %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_bal_decontam %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
sr_lmer_ns_decontam_kbl  <- sr_lmer_ns_decontam %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_ns_decontam %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        

sr_lmer_spt_decontam_kbl <- sr_lmer_spt_decontam %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_spt_decontam %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]

sr_lmer_bal_tv_kbl <- sr_lmer_bal_tv %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_bal_tv %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        
sr_lmer_ns_tv_kbl  <- sr_lmer_ns_tv %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_ns_tv %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
                        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]


sr_lmer_spt_tv_kbl <- sr_lmer_spt_tv %>%
        summary() %>%
        .$coefficients %>%
        data.frame(check.names = F) %>% 
        mutate(` ` = case_when(abs(`Pr(>|t|)`) < 0.001 ~ "***",
                               abs(`Pr(>|t|)`) < 0.01 ~ "**",
                               abs(`Pr(>|t|)`) < 0.05 ~ "*",
                               .default = " ")) %>% 
        rename("<i>p</i>-value" = "Pr(>|t|)",
               t_value = "t value") %>%
        merge(., 
              sr_lmer_spt_tv %>% confint() %>%
                        round(digits = 1) %>% 
                        format(nsmall = 1) %>%
                        as.data.frame() %>%
                        mutate(conf = paste("(",
                                            gsub(" ","", .[,1]),
                                            ", ",
                                            gsub(" ","", .[,2]),
                                            ")",
                                            sep = "")),
              by = 0
              ) %>%
        column_to_rownames("Row.names") %>%
        rownames_to_column(var = "x") %>% mutate(x = gsub("treatment|sample_type", "", x)) %>% mutate(x = gsub(":", " * ", x)) %>%
        mutate(x = gsub("bal_log_centered_final_reads", "log<sub>10</sub>(Final reads)", x)) %>%
        column_to_rownames(var = "x") %>% 
        mutate("Effect size (95% CI)" = 
                       paste(round(Estimate, digits = 1) %>%
                                     format(nsmall = 1),
                             conf, 
                             sep = " "),
               "<i>p</i>-value" = round(`<i>p</i>-value`, 3)) %>%
        dplyr::select(c("Effect size (95% CI)", "<i>p</i>-value", " ")) %>%
        .[c("(Intercept)",
                  "lyPMA",
                  "Benzonase",
                  "HostZERO",
                  "MolYsis",
                  "QIAamp"),]
        


tableS11 <- cbind(
        cbind(sr_lmer_bal_decontam_kbl, sr_lmer_ns_decontam_kbl, sr_lmer_spt_decontam_kbl),
        cbind(sr_lmer_bal_tv_kbl, sr_lmer_ns_tv_kbl, sr_lmer_spt_tv_kbl) %>% remove_rownames()) %>%
    kbl(format = "html", escape = 0) %>%
        add_header_above(c(" "= 1, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3, "BAL" = 3, "Nasal swab" = 3,"Sputum"= 3)) %>% 
        add_header_above(c(" " = 1, "Decontaminated data 1" = 9, "Decontaminated data 2"= 9)) %>% 
        kable_styling(full_width = 0, html_font = "sans")


tableS11

save_kable(tableS11, file = "Project_SICAS2_microbiome/7_Manuscripts/2022_MGK_Host_Depletion/Figures/tableS11.html", self_contained = T)

Beta diversity plot

figd2a <- ordinate(subset_samples(phyloseq_rel_nz, sample_type != "Neg." & sample_type != "Mock"), method = "PCoA", distance = "horn") %>%
        plot_ordination(phyloseq_rel_nz, ., col = "treatment") +
        #scale_color_viridis(discrete = 6, name = "Treatment", labels = c("Mock theoretical", "Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                           name = "Treatment",
                           breaks = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
                           labels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        #scale_shape(name = "Sample type", labels = c("Mock theoretical", "Mock")) +
        geom_point(size = 3) +
        theme_classic (base_size = 12, base_family = "sans") +
        facet_wrap(~sample_type, scales = "free") +
        labs(tag = "A") +
        ggtitle("Prevalence & abundance filtered data") +
        theme(plot.tag = element_text(size = 15), legend.position = "top")# +
        #stat_ellipse(type = "norm") +
        #stat_ellipse(type = "t")


figd2b <- ordinate(subset_samples(phyloseq_decontam$phyloseq_rel, sample_type != "Neg." & sample_type != "Mock" &
                                          S.obs != 0), method = "PCoA", distance = "horn") %>%
        plot_ordination(phyloseq_decontam$phyloseq_rel, ., col = "treatment") +
        #scale_color_viridis(discrete = 6, name = "Treatment", labels = c("Mock theoretical", "Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                           name = "Treatment",
                           breaks = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
                           labels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        #scale_shape(name = "Sample type", labels = c("Mock theoretical", "Mock")) +
        geom_point(size = 3) +
        theme_classic (base_size = 12, base_family = "sans") +
        facet_wrap(~sample_type, scales = "free") +
        labs(tag = "B") +
        ggtitle("Decontaminated data 1") +
        theme(plot.tag = element_text(size = 15), legend.position = "top")# +
        #stat_ellipse(type = "norm") +
        #stat_ellipse(type = "t")


figd2c <- ordinate(subset_samples(phyloseq_tv, sample_type != "Neg." & sample_type != "Mock"), method = "PCoA", distance = "horn") %>%
        plot_ordination(phyloseq_tv, ., col = "treatment") +
        #scale_color_viridis(discrete = 6, name = "Treatment", labels = c("Mock theoretical", "Control","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) +
        scale_color_manual(values = c("#e31a1c", "#fb9a99", "#33a02c", "#b2df8a", "#1f78b4", "#a6cee3"),
                           name = "Treatment",
                           breaks = c("Untreated","lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp"),
                           labels = c("Untreated", "lyPMA", "Benzonase", "HostZERO", "MolYsis", "QIAamp")) + #color using https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=6
        #scale_shape(name = "Sample type", labels = c("Mock theoretical", "Mock")) +
        geom_point(size = 3) +
        theme_classic (base_size = 12, base_family = "sans") +
        facet_wrap(~sample_type, scales = "free") +
        labs(tag = "C") +
        ggtitle("Decontaminated data 2") +
        theme(plot.tag = element_text(size = 15), legend.position = "top")# +
        #stat_ellipse(type = "norm") +
        #stat_ellipse(type = "t")


figA3 <- ggarrange(figd2a, figd2b, figd2c, common.legend = T, nrow = 3)
figA3

List of contaminants (Tinyvamp)

#Stratified by sample type

prev_neg

prev_all

sample_data(phyloseq_unfiltered$phyloseq_rel)$is.neg <- grepl("Neg", sample_data(phyloseq_unfiltered$phyloseq_rel)$sample_type)

contaminants_tv <- data.frame(
        Taxa = subset(taxa_names(phyloseq_unfiltered$phyloseq_count),
       !(taxa_names(phyloseq_unfiltered$phyloseq_count) %in%
               taxa_names(phyloseq_tv))
)) 
       


merged_contaminants <- merge(contaminants_tv, prev_all %>% rownames_to_column("Taxa"), by = "Taxa") %>%
        merge(., prev_neg %>% rownames_to_column("Taxa"), by = "Taxa") %>%
        dplyr::select(c("Taxa", "Prevalence (all)", "Prevalence (negative controls)")) %>%
        .[order(-.$"Prevalence (all)", -.$"Prevalence (negative controls)"),] %>%
        remove_rownames() %>%
        subset(., .$"Prevalence (all)" != 0)


tableA3 <- merged_contaminants %>% 
        mutate(Taxa = species_italic2(Taxa)) %>%
        kbl(format = "html", escape = F) %>%
        kable_styling(full_width = 0, html_font = "sans") 

tableA3